tag:blogger.com,1999:blog-43846928367099031462024-03-18T04:38:04.842-07:00Dispatch from the Digital Health FrontierAs president of the Mayo Clinic Platform, I lead a portfolio of new digital platform businesses focused on transforming health by leveraging artificial intelligence, the internet of things, and an ecosystem of partners for Mayo Clinic. This is made possible by an extraordinary team of people at Mayo and collaborators worldwide. This blog will document their story.
John Halamkahttp://www.blogger.com/profile/04550236129132159307noreply@blogger.comBlogger1879125tag:blogger.com,1999:blog-4384692836709903146.post-28973475551968688462021-11-09T11:54:00.003-08:002021-11-09T11:54:41.031-08:00We Moved!<p class="xmsonormal">Thank you for reading our blog. Paul and I will continue
to publish "Dispatch from the Digital Health Frontier" on <a href="https://www.mayoclinicplatform.org/blog/">https://www.mayoclinicplatform.org/blog/</a>.<o:p></o:p></p><p><br /></p>John Halamkahttp://www.blogger.com/profile/04550236129132159307noreply@blogger.com0tag:blogger.com,1999:blog-4384692836709903146.post-57424427609818212662021-10-21T09:18:00.000-07:002021-10-21T09:18:48.990-07:00Machine Learning Can Make Lab Testing More Precise<h4 style="text-align: left;">An analysis of over 2 billion lab
test results suggests a deep learning model can help create personalized
reference ranges, which in turn would enable clinicians to monitor health and
disease better.</h4><div class="separator" style="clear: both; text-align: center;"><a href="https://1.bp.blogspot.com/-8HU0F_cYxOg/YW27RY0--nI/AAAAAAAALUE/Y_yTeKyZNHcEezIX5KD__2eAOkERNNSwACLcBGAsYHQ/s800/WF516211_0122.jpg" imageanchor="1" style="margin-left: 1em; margin-right: 1em;"><img border="0" data-original-height="533" data-original-width="800" height="266" src="https://1.bp.blogspot.com/-8HU0F_cYxOg/YW27RY0--nI/AAAAAAAALUE/Y_yTeKyZNHcEezIX5KD__2eAOkERNNSwACLcBGAsYHQ/w400-h266/WF516211_0122.jpg" width="400" /></a></div><p class="MsoNormal" style="line-height: 150%;"><o:p></o:p></p>
<p class="MsoNormal" style="line-height: 150%;"><i>Paul Cerrato, MA, senior research
analyst and communications specialist, Mayo Clinic Platform and John Halamka,
M.D., president, Mayo Clinic Platform, wrote this article.</i></p><p class="MsoNormal" style="line-height: 150%;"><o:p></o:p></p>
<p class="MsoNormal" style="line-height: 150%;">Almost every patient has blood
drawn to measure a variety of metabolic markers. Typically, test results come
back as a numeric or text value accompanied by a reference range which
represents normal values. If total serum cholesterol level is below 200 mg/dl
or serum thyroid hormone level is 4.5 to 12.0 mcg/dl, clinicians and patients
assume all is well. But suppose Helen’s safe zone varies significantly from
Mary’s safe zone. If that were the case, it would suggest a one-size-fits-all
reference range misrepresents an individual’s health status. That position is
supported by studies that found the distribution of more than half of all lab
test results, which rely on standard reference ranges, differ when personal
characteristics are considered.<sup>1<o:p></o:p></sup></p><p class="MsoNormal" style="line-height: 150%;">With these concerns in mind,
Israeli investigators from the Weismann Institute and Tel Aviv Sourasky Medical
Center extracted data on 2.1 billion lab measurements from EHR records, taken
from 2.8 million adults for 92 different lab tests. Their goal was to create
“data-driven reference ranges that consider age, sex, ethnicity, disease status,
and other relevant characteristics.”<sup>1 </sup>To accomplish that goal, they used
machine learning and computational modeling to segment patients into different
“bins'' based on health status, medication intake, and chronic disease.<sup>2. </sup>That
in turn left the team with about half a billion lab results from the initial
2.8 million people, which they used to model a set of reference lab values that
more precisely reflected the ranges of healthy persons. Those ranges could then
be used to predict patients’ “future lab abnormalities and subsequent disease.”</p><p class="MsoNormal" style="line-height: 150%;"><o:p></o:p></p><p class="MsoNormal" style="line-height: 150%;">Taking their investigation one step
forward, Cohen et al. used their new algorithms to evaluate the risk of
specific disorders amongst healthy individuals. When they looked at anemia cut
offs like hemoglobin and mean corpuscular volume, a measurement of red blood
cell size, their newly created risk calculators were able to separate anemic
patients into groups at high risk for microcytic and macrocytic anemia from
those with a risk no higher than the average nonanemic population. Similar
benefits were observed when the researchers applied their models to
prediabetes: “…using a personalized risk model, we can improve the
classification of patients who are prediabetic and identify patients at risk 2
years earlier compared to classification based merely on current glucose
levels.”</p><p class="MsoNormal" style="line-height: 150%;"><o:p></o:p></p><p class="MsoNormal" style="line-height: 150%;">William Morice, M.D., Ph.D., chair
of the Department of Laboratory Medicine and Pathology (DLMP) at Mayo Clinic
and president of Mayo Clinic Laboratories, immediately saw the value of this
type of data analysis: “In the ‘era of big data and analytics,’ it is almost
unconscionable that we still use ‘normal reference ranges’ that lack contextual
data, and possibly statistical power, to guide clinicians in the clinical
interpretation of quantitative lab results. I was taught this by <a href="https://www.mayo.edu/research/faculty/rinaldo-piero-m-d-ph-d/bio-00027619">Dr.
Piero Rinaldo</a>, a medical geneticist in our department and a pioneer in this
field, who focuses on its application to screening for inborn errors of
metabolism. He has developed an elegant tool that is now used globally for this
application, Collaborative Laboratory Integrated Reports (CLIR).”</p><p class="MsoNormal" style="line-height: 150%;"><o:p></o:p></p><p class="MsoNormal" style="line-height: 150%;">
</p><p class="MsoNormal" style="line-height: 150%;">During a recent conversation with
Piero Rinaldo, M.D., Ph.D., he explained that Mayo Clinic has been using a more
personalized approach to lab testing since 2015 and stated that “CLIR is a
shovel-ready software for the creation of collaborative precision reference
ranges.” The web-based application has been used to create several personalized
data sets that can improve clinicians’ interpretation of lab test results. It
has been deployed by Dr. Rinaldo and his associates to improve the screening of
newborns for congenital hyperthyroidism.<sup>3. </sup>The software performs
multivariate pattern recognition on lab values collected from 7 programs,
including more than 1.9 million lab test results. CLIR is able to integrate
covariate-adjusted results of different tests into a set of customized
interpretive tools that physicians can use to better distinguish between false
positive and true positive test results.<o:p></o:p></p><p class="MsoNormal" style="line-height: 150%;"><br /></p><p class="MsoNormal" style="line-height: 150%;"><b>References</b></p>
<p class="MsoNormal" style="line-height: 150%;">1. Tang
A, Oskotsky T, Sirota M. Personalizing routine lab tests with machine Learning. <i>Nature Medicine. </i>2021; 27:1510-1517.</p><p class="MsoNormal" style="line-height: 150%;"><o:p></o:p></p>
<p class="MsoNormal" style="line-height: 150%;">2. Cohen
N, Schwartzman O, Jaschek R et al. Personalized lab test models to quantify
disease potentials in healthy individuals. <i style="mso-bidi-font-style: normal;">Nature
Medicine.</i>2021; 27: 1582-1591. <o:p></o:p></p>
<p class="MsoNormal" style="line-height: 150%;">3. Rowe
AD, Stoway SD, Ahlman H et al. A Novel Approach to Improve Newborn Screening
for Congenital Hypothyroidism by Integrating Covariate-Adjusted Results of
Different Tests into CLIR Customized Interpretive Tools. <i style="mso-bidi-font-style: normal;">Inter J Neonatal Screening. </i>2021. 7:23 <a href="https://doi.org/10.3390/ijns7020023"><span style="color: #0563c1;">https://doi.org/10.3390/ijns7020023</span></a><o:p></o:p></p>John Halamkahttp://www.blogger.com/profile/04550236129132159307noreply@blogger.com0tag:blogger.com,1999:blog-4384692836709903146.post-51708297317895764142021-10-13T08:44:00.000-07:002021-10-13T08:44:20.354-07:00Gastroenterology Embraces Artificial Intelligence <h4 style="text-align: left;">AI and machine learning have the potential to redefine
the management of several GI disorders.</h4><div class="separator" style="clear: both; text-align: center;"><a href="https://1.bp.blogspot.com/-mp27ETRRhLE/YWb9uYR6y3I/AAAAAAAALRk/j4WuNITVIxU5SSd_2MPnje79WhEhDJ38wCLcBGAsYHQ/s800/shutterstock_1452312035.jpg" imageanchor="1" style="margin-left: 1em; margin-right: 1em;"><img border="0" data-original-height="514" data-original-width="800" height="258" src="https://1.bp.blogspot.com/-mp27ETRRhLE/YWb9uYR6y3I/AAAAAAAALRk/j4WuNITVIxU5SSd_2MPnje79WhEhDJ38wCLcBGAsYHQ/w400-h258/shutterstock_1452312035.jpg" width="400" /></a></div><br /><div><i>John Halamka, M.D., president, Mayo Clinic Platform, and
Paul Cerrato, senior research analyst and communications specialist, Mayo
Clinic Platform, wrote this article.</i></div>
<p class="MsoNormal" style="line-height: 150%;">Colonoscopy is one of the true success stories in modern
medicine. <a href="https://pubmed.ncbi.nlm.nih.gov/23784448/">Studies</a> have demonstrated that colonoscopy
screening detects the cancer at a much earlier stage, reducing the risk of
invasive tumors and metastatic disease, and <a href="https://pubmed.ncbi.nlm.nih.gov/29532085/">reducing mortality</a>. However, while colorectal cancer is highly
preventable, it is the <a href="https://www.cdc.gov/cancer/colorectal/statistics/index.htm">third leading
cause of cancer-related deaths</a>
in the U.S. About 148,000 individuals develop the malignancy and over 53,000
die from it each year. We asked ourselves a question: can AI improve the
detection of this and related gastrointestinal disorders?</p>
<p class="MsoNormal" style="line-height: 150%;">As we explained in <a href="https://www.routledge.com/The-Digital-Reconstruction-of-Healthcare-Transitioning-from-Brick-and-Mortar/Cerrato-Halamka/p/book/9780367555979"><i>The
Digital Reconstruction of Healthcare</i></a><i>, </i>one of the challenges in making an accurate diagnosis of GI disease is
differentiating between disorders that look similar at the cellular level. For
example, because environmental enteropathy and celiac disease overlap histopathologically,
deep learning algorithms have been designed to analyze biopsy slides to detect
the subtle differences between the two conditions. Syed et al.<sup>1</sup> used
a combination of convolutional and deconvolutional neural networks in a
prospective analysis of over 3,000 biopsy images from 102 children. They were
able to tell the differences between environmental enteropathy, celiac disease,
and normal controls with an accuracy rating of 93.4%, and a false negative rate
of 2.4%. Most of these mistakes occurred when comparing celiac patients to
healthy controls.</p>
<p class="MsoNormal" style="line-height: 150%;">The investigators also identified several biomarkers that
may help separate the two GI disorders: interleukin 9, interleukin 6,
interleukin 1b, and interferon-induced protein 10 were all helpful in making an
accurate prediction regarding the correct diagnosis. The potential benefits to
this deep learning approach become obvious when one considers the arduous
process that patients have to endure to reach a definitive diagnosis of either
disorder: typically, they must undergo 4 to 6 biopsies and may need several
endoscopic procedures to sample various sections of the intestinal tract
because the disorder may affect only specific areas along the lining and leave
other areas intact.</p>
<p class="MsoNormal" style="line-height: 150%;">Several randomized controlled trials have been conducted
to support the use of ML in gastroenterology. Chinese investigators, working in
conjunction with Beth Israel Deaconess Medical Center and Harvard Medical
School, tested a convolutional neural network to determine if it was capable of
improving the detection of precancerous colorectal polyps in real time.<sup>2</sup>
The need for a better system of detecting these growths is evident, given the
fact that more than 1 in 4 adenomas are missed during coloscopies. To address
the problem, Wang et al. randomized more than 500 patients to routine
colonoscopy and more than 500 to computer-assisted colonoscopies. In the final
analysis, the adenoma detection rate (ADR) was higher in the ML-assisted group
(29.1% vs. 20.3%, P < 0.001). The higher ADR occurred because the algorithm
was capable of detecting a greater number of smaller adenomas (185 vs. 102).
There were no significant differences in the detection of large polyps.</p>
<p class="MsoNormal" style="line-height: 150%;">Nayantara Coelho-Prabhu, M.D., a gastroenterologist at Mayo
Clinic, points out, however, that the clinical relevance of detection of
diminutive polyps remains to be determined. “Yet, there is definite clinical
importance in the subsequent development of computer assisted diagnosis (CADx) or
polyp characterization algorithms. These will help clinicians determine
clinically relevant polyps, and possibly advance the resect and discard
practice. It also will help clinicians adequately assess margins of polyps, so
that complete removal can be achieved, thus decreasing future recurrences.”</p>
<p class="MsoNormal" style="line-height: 150%;">Randomized clinical trials demonstrated that a convolutional
neural network in combination with deep reinforcement learning (collectively
called the WISENSE system) can reduce the number of blind spots during
endoscopy intended to evaluate the esophagus, stomach, and duodenum in real
time. “A total of 324 patients were recruited and randomized; 153 and 150
patients were analysed in the WISENSE and control group, respectively. Blind
spot rate was lower in WISENSE group compared with the control (5.86% vs
22.46%, p<0.001) . . .”<sup>3</sup></p>
<p class="MsoNormal" style="line-height: 150%;">Mayo Clinic’s Endoscopy Center, utilizing Mayo Clinic
Platform’s resources, has also been exploring the value of machine learning in
GI care with the assistance of Endonet, a comprehensive library of endoscopic
videos and images, linked to clinical data including symptoms, diagnoses,
pathology, and radiology. These data will include unedited full-length videos as
well as video summaries of the procedure including landmarks, specific abnormalities,
and anatomical identifiers. Dr. Coelho-Prabhu explains that the idea is to have
different user interfaces: </p>
<p class="MsoNormal" style="line-height: 150%;">“From the patient’s perspective, it will serve as an
electronic video record of all their procedures, and future procedures can be
tailored to survey prior abnormal areas as needed.</p>
<p class="MsoNormal" style="line-height: 150%;"><span style="color: black; mso-themecolor: text1;">From a research perspective, this will be a diverse and
rich library including large volumes of specialized populations such as
Barrett’s esophagus, inflammatory bowel disease, familial polyposis syndromes.
The additional strength is that Mayo Clinic provides highly specialized care,
especially to these select populations. We can develop AI algorithms to advance
medical care using this library. From a hospital system perspective, this would
serve as a reference library, guiding endoscopists, including for advanced
therapeutic procedures in the future. It also could be used to measure and
monitor quality indicators in endoscopy. From an educational standpoint, this
library can be developed into a teaching set for both trainee and advanced
practitioners looking for CME opportunities. From industry perspective, this
database could be used to train/validate commercial AI algorithms.”<o:p></o:p></span></p>
<p class="MsoNormal" style="line-height: 150%;">AI and machine learning may not be the panacea some
technology enthusiasts imagine it to be, but there’s little doubt they are
becoming an important partner in the road to more personalized patient care.</p>
<p class="MsoNormal" style="line-height: 150%;"><b><br /></b></p><p class="MsoNormal" style="line-height: 150%;"><b>References</b></p><p class="MsoNormal" style="line-height: 150%;"><span style="line-height: 150%; text-indent: -0.5in;"><span style="font-family: inherit;">1. Syed
S, Al-Bone M, Khan MN, et al. Assessment of machine learning detection of
environmental enteropathy and celiac disease in children. <i>JAMA Network Open.
</i>2019;2:e195822.</span></span></p><p class="MsoNormal" style="line-height: 150%;"><span style="text-indent: -0.5in;"><span style="font-family: inherit;">2. Wang
P, Berzin TM, Brown JR, et al. Real-time automatic detection system increases colonoscopic
polyp and adenoma detection rates: a prospective randomised controlled study.
Gut. 2019;68:1813–1819.</span></span></p>
<p class="MsoNormal" style="line-height: 150%;"><span style="color: black; mso-themecolor: text1;"><span style="font-family: inherit;">3. Wu L,
Zhang J, Zhou W, et al Randomised controlled trial of WISENSE, a real-time
quality improving system for monitoring blind spots during esophagogastroduodenoscopy.
Gut. 2019;68:2161–2169.</span></span></p>John Halamkahttp://www.blogger.com/profile/04550236129132159307noreply@blogger.com0tag:blogger.com,1999:blog-4384692836709903146.post-78792297474907517242021-10-06T14:06:00.000-07:002021-10-06T14:06:56.940-07:00Societal Resilience Requires a Public Health Focus<h4 style="text-align: left;">We must make a serious commitment
to increase financial resources and provide better analytics for real world
evidence/real time data in support of public health.</h4><div class="separator" style="clear: both; text-align: center;"><a href="https://1.bp.blogspot.com/-LTnDCA8U4TI/YVyolncegxI/AAAAAAAALPk/4-S0OP2sJrgX_CD9E1i34hrkpD88UOjsQCLcBGAsYHQ/s800/shutterstock_623235479.jpg" imageanchor="1" style="margin-left: 1em; margin-right: 1em;"><img border="0" data-original-height="422" data-original-width="800" height="211" src="https://1.bp.blogspot.com/-LTnDCA8U4TI/YVyolncegxI/AAAAAAAALPk/4-S0OP2sJrgX_CD9E1i34hrkpD88UOjsQCLcBGAsYHQ/w400-h211/shutterstock_623235479.jpg" width="400" /></a></div><br /><div><i>John Halamka, M.D., president, Mayo
Clinic Platform, and Paul Cerrato, senior research analyst and communications
specialist, Mayo Clinic Platform, wrote this article.</i></div><p class="MsoNormal" style="line-height: 150%;"><o:p></o:p></p>
<p class="MsoNormal" style="line-height: 150%;">Public health has been underfunded
for decades. That neglect has had a profound impact since the COVID-19 pandemic
has taken hold, and awakened policy makers and thought leaders to the need for
more investment.</p><p class="MsoNormal" style="line-height: 150%;"><o:p></o:p></p>
<p class="MsoNormal" style="line-height: 150%;">Consider the statistics: The U.S.
spends about $3.6 trillion each year on health <a href="https://www.tfah.org/report-details/publichealthfunding2020/">but less
than 3%</a> of that amount on public health and prevention. A <a href="https://www.forbes.com/sites/williamhaseltine/2020/10/21/underfunding-public-health-harms-americans-beyond-covid-19/?sh=3a667ae9419c">2020
Forbes report</a> likewise pointed out that “From the late 1960s to the 2010s,
the federal share of total health expenditure for public health dropped from 45
percent to 15 percent.” This relative indifference to public health is partly
responsible for the nation’s mixed response to the SARS-CoV-2 pandemic. A <a href="https://www.mckinsey.com/industries/public-and-social-sector/our-insights/not-the-last-pandemic-investing-now-to-reimagine-public-health-systems">recent
McKinsey & Company</a> analysis concluded: “Government leaders remain
focused on navigating the current crisis, but making smart investments now can
both enhance the ongoing COVID-19 response and strengthen public-health systems
to reduce the chance of future pandemics. Investments in public health and
other public goods are sorely undervalued; investments in preventive measures,
whose success is invisible, even more so.”</p><p class="MsoNormal" style="line-height: 150%;"><o:p></o:p></p>
<p class="MsoNormal" style="line-height: 150%;">Among the other “public goods” that
require more investment is population health management and analytics. Although
experts continue to debate the differences between public health and population
health, most are unimportant. For our purposes, population health refers to the
status of a specific group of individuals, whether they reside in a specific
city, state, or country. Public health usually casts a wider net, concerned
about the status of the entire population. Managing the health of these
subgroups requires an analytical approach that can take into account a long
list of variables, including social determinants of health (SDoH), the content
of their medical records, and much more. SDoH data from Change Health care, for
instance, has demonstrated that economic stability index (ESI) is a strong
predictor of health care utilization. ESI is a cluster model that uses market
behavior and financial attitudes o group individuals into one of 30 categories,
with category 1 representing persons most likely to be economically stable and
category 30 least likely to be stable. The figure, which links race, ESI and health
care utilization in Kentucky, suggests that Blacks/African Americans are far
less likely to be economically stable (category 1). The same analysis found
that Blacks/African Americans were almost twice as likely to use the ED
compared to Whites (30.5% vs 18.1%). A growing number of health care
organizations are starting to see the value of such population health metrics
and are incorporating these statistics into their decision making.</p><p class="MsoNormal" style="line-height: 150%;"><o:p></o:p></p>
<p class="MsoNormal" style="line-height: 150%;">Among the valuable sources of data
that can inform population health are patient surveys, clinical registries, and
EHRs. Several traditional analytics tools are available to extract actionable
insights from these data sources, including logistic regression. Over the
decades, several major studies have also generated risk scoring systems to
improve public health. The Framingham heart health risk score has been used for
many years to assess the likelihood of developing cardiovascular disease over a
10-year period. Because the scoring system can help predict the onset of heart
disease, it can also serve as a useful tool in creating population-based
preventive programs to reduce that risk. The tool requires patients to provide
their age, gender, smoking status, total cholesterol, HDL cholesterol, systolic
blood pressure, and whether they are taking antihypertensive medication. The
American Diabetes Association has developed its own risk scoring method to
assess the likelihood of type 2 diabetes in the population. The tool takes into
account age, gender, history of gestational diabetes, physical activity level,
family history of diabetes, hypertension, height and weight. Another analytics
methodology that has value in population health is the LACE Index. The acronym
stands for length of stay, acuity of admission, Charlson comorbidity index (CCI),
and number of emergency department visits in the preceding 6 months. More
recently, there are several AI-based analytic tools currently being used to
improve population health. A <a href="https://bmjopen.bmj.com/content/10/10/e037860">review of ML-related
analytic methods</a> found that neural networks based algorithms are the most
commonly used (41%) in this context, compared to 25.5% for support vector
machines, and 21% for random forest modeling.</p><p class="MsoNormal" style="line-height: 150%;"><o:p></o:p></p>
<p class="MsoNormal" style="line-height: 150%;">There is no way of knowing how the
world would have coped with COVID-19 had policy makers fully invested in public
and population health programs and analytics. But there’s little doubt that we’ll
all fare much better during the next health crisis if we put more time, energy,
and resources into these initiatives.</p><p class="MsoNormal" style="line-height: 150%;"><o:p></o:p></p>John Halamkahttp://www.blogger.com/profile/04550236129132159307noreply@blogger.com0tag:blogger.com,1999:blog-4384692836709903146.post-31867032837390501682021-09-30T09:06:00.000-07:002021-09-30T09:06:26.471-07:00Reimagining the FDA’s Role in Digital Medicine<h4 style="text-align: left;">In addition to evaluating the
safety of software as a medical device (SaMD), the agency needs to devote more
resources to evaluating its efficacy and quality.</h4><div class="separator" style="clear: both; text-align: center;"><a href="https://1.bp.blogspot.com/-y8rM9k6lpQY/YU4rlX9T1MI/AAAAAAAALLM/7Q_S4VKpWI0R3lGjUDAOLQ0qXiHcszIFQCLcBGAsYHQ/s800/FDA.jpg" imageanchor="1" style="margin-left: 1em; margin-right: 1em;"><img border="0" data-original-height="489" data-original-width="800" height="245" src="https://1.bp.blogspot.com/-y8rM9k6lpQY/YU4rlX9T1MI/AAAAAAAALLM/7Q_S4VKpWI0R3lGjUDAOLQ0qXiHcszIFQCLcBGAsYHQ/w400-h245/FDA.jpg" width="400" /></a></div><p class="MsoNormal" style="line-height: 150%;"><o:p></o:p></p>
<p class="MsoNormal" style="line-height: 150%;"><i>John Halamka, M.D., president, Mayo
Clinic Platform, and Paul Cerrato, senior research analyst and communications
specialist, Mayo Clinic Platform, wrote this article.</i></p><p class="MsoNormal" style="line-height: 150%;"><o:p></o:p></p>
<p class="MsoNormal" style="line-height: 150%;">The FDA’s approach to software as a
medical device (SaMD) has been evolving. Consider a few examples.</p><p class="MsoNormal" style="line-height: 150%;"><o:p></o:p></p>
<p class="MsoNormal" style="line-height: 150%;">In 2018, IDx-DR, a software system used
to improve screening for retinopathy, a common complication of diabetes that
affects the eye, became the first AI-based medical device to receive <a href="https://www.fda.gov/news-events/press-announcements/fda-permits-marketing-artificial-intelligence-based-device-detect-certain-diabetes-related-eye">US
Food and Drug Administration clearance</a> to “detect greater than a mild level
of … diabetic retinopathy in adults who have diabetes.” To arrive at that
decision, the agency not only reviewed data to establish its safety, it also
took into account prospective studies, an essential form of evidence that
clinicians look for when trying to decide if a device or product is worth
using. The software was the first medical device approved by the FDA that does
not require the services of a specialist to interpret the results, making it a
useful tool for health care providers who may not normally be involved in eye
care. The FDA clearance emphasized the fact that IDx-DR is a screening tool not
a diagnostic tool, stating that patients with positive results should be
referred to an eye care professional. The algorithm built into the IDx-DR
system is intended to be used with the Topcon NW400 retinal camera and a cloud
server that contains the software.</p><p class="MsoNormal" style="line-height: 150%;"><o:p></o:p></p>
<p class="MsoNormal" style="line-height: 150%;">Similarly, FDA looked at a
randomized prospective trial before approval of a machine learning-based
algorithm that can help endoscopists improve their ability to detected smaller,
easily missed colonic polyps. Its recent clearance of GI Genius by Medtronic
was based on a clinical trial published in <a href="https://pubmed.ncbi.nlm.nih.gov/32371116/"><i>Gastroenterology</i></a><i>, </i>in which investigators in Italy evaluated data from 685 patients,
comparing a group that underwent the procedure with the help of the computer-aided detection (CADe) system to a group who acted as controls. Repici et al
found that the adenoma detection rate was significantly higher in the CADe
group, as was the detection rate for polyps 5 mm or smaller, which led to the
conclusion: “Including CADe in colonoscopy examinations increases detection of
adenomas without affecting safety.”</p><p class="MsoNormal" style="line-height: 150%;"><o:p></o:p></p>
<p class="MsoNormal" style="line-height: 150%;">Their findings raise several
questions: is it reasonable to assume that a study of 600+ Italians would apply
to a U.S. population, which has different demographic characteristics? More
importantly, were the 685 patients representative of the general public, including
adequate numbers of persons of color and those in lower socioeconomic groups?
While the <i>Gastroenterology </i>study did report enough female patients,
there is no mention of these other marginalized groups. </p><p class="MsoNormal" style="line-height: 150%;"><o:p></o:p></p>
<p class="MsoNormal" style="line-height: 150%;"><a href="https://pubmed.ncbi.nlm.nih.gov/33820998/">An independent 2021 analysis</a>
of FDA approvals has likewise raised several concerns about the effectiveness
and equity of several recently approved AI algorithms. Eric Wu from Stanford
University and his colleagues examined the FDA’s clearance of 130 devices and
found the vast majority were approved based on retrospective studies (126 of
130). And when they separated all 130 devices into low- and high-risk subgroups
using FDA guidelines, they found none of the 54 high-risk devices had been evaluated
by prospective trials. Other shortcomings documented in Wu’s analysis included
the following:</p><p class="MsoNormal" style="line-height: 150%;"></p><ul style="text-align: left;"><li><span style="text-indent: -0.25in;"><span style="font-family: inherit;">Of the 130 approved
products, 93 did not report multi-site evaluation.</span></span></li><li><span style="text-indent: -0.25in;"><span style="font-family: inherit;">Fifty-nine of the approved AI
devices included no mention of the sample size of the test population. </span></span></li><li><span style="text-indent: -0.25in;"><span style="font-family: inherit;">Only 17 of the approved
devices discussed a demographic subgroup. </span></span></li></ul><p></p>
<p class="MsoNormal" style="line-height: 150%;">We would certainly like to see the FDA take a more
thorough approach to AI-based algorithm clearance, but in lieu of that, several
leading academic medical centers, including Mayo Clinic, are contemplating a
more holistic and comprehensive approach to algorithmic evaluation. It would include
establishing a standard labeling schema to document the characteristics,
behavior, efficacy, and equity of AI systems, to reveal the properties of
systems necessary for stakeholders to assess them and build the trust necessary
for safe adoption. The schema will also support assessment of the portability
of systems to disparate datasets. The labeling schema will serve as an
organizational framework that specifies the elements of the label. Label
content will be specified in sections that will likely include:</p><p class="MsoNormal" style="line-height: 150%;"></p><ul style="text-align: left;"><li>model
details such as name, developer, date of release, and version,</li><li>the
intended use of the system,</li><li>performance
measures,</li><li>accuracy
metrics, and</li><li>training
data and evaluation data characteristics</li></ul><p></p><p class="MsoNormal" style="line-height: 150%;"><o:p></o:p></p>
<p class="MsoNormal" style="line-height: 150%;"><o:p></o:p></p>
<p class="MsoNormal" style="line-height: 150%;"><o:p></o:p></p>
<p class="MsoNormal" style="line-height: 150%;"><o:p></o:p></p>
<p class="MsoNormal" style="line-height: 150%;">While it makes no sense to
sacrifice the good in pursuit of the perfect, the current regulatory framework
for evaluating SaMD is far from perfect. Combining a more robust FDA approval
process with the expertise of the world’s leading medical centers will offer
our patients the best of both worlds.</p>John Halamkahttp://www.blogger.com/profile/04550236129132159307noreply@blogger.com0tag:blogger.com,1999:blog-4384692836709903146.post-37864825682318057522021-09-09T13:21:00.003-07:002021-09-09T13:21:26.220-07:00Secure Computing Enclaves Move Digital Medicine Forward<h4 style="text-align: left;">By providing a safe, secure environment, novel approaches
enable health care innovators to share data without opening the door to snoopers
and thieves.</h4><div class="separator" style="clear: both; text-align: center;"><a href="https://1.bp.blogspot.com/-WEXIG_Vp1Qk/YTpsSDzgM5I/AAAAAAAALFo/s7Fgh_NpMWA4Q3uvPo5_7vCTiFG377REQCLcBGAsYHQ/s800/shutterstock_347970266.jpg" imageanchor="1" style="margin-left: 1em; margin-right: 1em;"><img border="0" data-original-height="497" data-original-width="800" height="249" src="https://1.bp.blogspot.com/-WEXIG_Vp1Qk/YTpsSDzgM5I/AAAAAAAALFo/s7Fgh_NpMWA4Q3uvPo5_7vCTiFG377REQCLcBGAsYHQ/w400-h249/shutterstock_347970266.jpg" width="400" /></a></div><p><i>John Halamka, M.D., president, Mayo Clinic Platform, and
Paul Cerrato, senior research analyst and communications specialist, Mayo
Clinic Platform, wrote this article.</i></p>
<p class="MsoNormal" style="line-height: 150%;"><span style="background-color: white;">We know that
bringing together AI algorithms and data in ways that preserve privacy and
intellectual property is one of the keys to delivering the next generation of clinical
decision support. But meeting that challenge requires health care innovators to look
to other innovators who themselves have created unique cybersecurity solutions.
Among these “Think outside the box” solutions are products and services from
vendors like TripleBlind, Verily, Beekeeper.AI/Microsoft, Terra, and Nvidia.</span></p>
<p class="MsoNormal" style="line-height: 150%;"><span style="background: white;">The concept of secure
computing enclaves has been around for many years. Apple created its secure
enclave, a subsystem built into its </span><a href="https://support.apple.com/guide/security/secure-enclave-sec59b0b31ff/web"><span style="background: white; mso-ascii-font-family: Calibri; mso-bidi-font-family: Calibri; mso-fareast-font-family: "Times New Roman"; mso-hansi-font-family: Calibri;">systems
on a chip (SoC),</span></a><span style="background: white;"> which in turn is “an
integrated circuit that incorporates multiple components into a single chip,”
including an application processor, secure enclave, and other coprocessors.
Apple explains that “The Secure Enclave is isolated from the main processor to
provide an extra layer of security and is designed to keep sensitive user data
secure even when the Application Processor kernel becomes compromised. It
follows the same design principles as the SoC does—a boot ROM to establish a
hardware root of trust, an AES [advanced encryption standard] engine for
efficient and secure cryptographic operations, and protected memory. Although
the Secure Enclave doesn’t include storage, it has a mechanism to store
information securely on attached storage separate from the NAND flash storage
that’s used by the Application Processor and operating system.” The secure
enclave is embedded into the latest versions of its iPhone, iPad, Mac
computers, Apple TV, Apple Watch, and Home Pod.</span></p>
<p class="MsoNormal" style="line-height: 150%;"><span style="background-color: white;">While this security
measure provides users when an extra layer of protection, because it’s a
hardware-based solution, its uses are limited. With that in mind, several
vendors have created software-based enclaves that are more readily adapted by
customers. At Mayo Clinic Platform, we are deploying TripleBlind’s services to
facilitate sharing data with our many external partners. It allows Mayo Clinic
to test its algorithms using another organization’s data without either party
losing control of its assets. Similarly, we can test an algorithm from one of our
academic or commercial partners with Mayo Clinic data, or test an outside
organization’s data with another outside organization’s data.</span></p>
<p class="MsoNormal" style="line-height: 150%;"><span style="background-color: white;">How is this “magic”
performed?</span> <span style="background-color: white;">Of course, it’s always about
the math. TripleBlind allows the use of distributed data that is accessed but
never moved or revealed; it always remains one-way encrypted with no decryption
possible. TripleBlind’s novel cryptographic</span> <span style="background-color: white;">approaches can operate on any type of data
(structured or unstructured images, text, voice, video), and perform any
operation, including training of and inferring from AI and ML algorithms. An
organization’s data remains fully encrypted throughout the transaction, which
means that a third party never sees the raw data because it is stored behind
the data owner organization’s firewall. In fact, there is no decryption key
available, ever.</span> <span style="background-color: white;">When two health care
organizations partner to share data, for instance, TripleBlind software de-identifies
their data via one-way encryption; then, both partners access each other’s
one-way encrypted data through an Application Programming Interface (API). That
means each partner can use the other’s data for training an algorithm, for
example, which in turn allows them to generate a more generalizable, less
biased algorithm. During a recent conversation with Riddhiman Das, CEO for
TripleBlind, he explained: “To build robust algorithms, you want to be able to
access diverse training data so that your model is accurate and can generalize
to many types of data. Historically, health care organizations have had to send
their data to one another to accomplish this goal, which creates unacceptable
risks. TripleBlind performs one-way encryption from both interacting
organizations, and because there is no decryption possible, you cannot
reconstruct the data. In addition, the data can only be used by an algorithm
for the specific purpose spelled out in the business agreement.”</span></p>
<p class="MsoNormal" style="line-height: 150%;"><span style="background-color: white;">Developing
innovative technological services is exciting work, with the potential to
reshape the health care ecosystem worldwide. But along with the excitement is
the challenge of keeping data safe and secure. Taking advantage of the many
secure computing enclaves available on the market allows us to do just that.</span></p>John Halamkahttp://www.blogger.com/profile/04550236129132159307noreply@blogger.com0tag:blogger.com,1999:blog-4384692836709903146.post-27238828408968617672021-08-31T11:54:00.000-07:002021-08-31T11:54:03.227-07:00Breast Cancer Screening: We Can Do Better<h4 style="text-align: left;">The three risk assessment tools now in use fall far
short. Using the latest deep learning techniques, investigators are developing
more personalized ways to locate women at high risk.</h4><div class="separator" style="clear: both; text-align: center;"><a href="https://1.bp.blogspot.com/-qu2HUszllsc/YS56k_XyrNI/AAAAAAAAK_k/MHazLyq5P-EY4cmcrfJtZBAxKjOCHSd4wCLcBGAsYHQ/s714/shutterstock_1207894222%2B%25283%2529.jpg" imageanchor="1" style="margin-left: 1em; margin-right: 1em;"><img border="0" data-original-height="393" data-original-width="714" height="220" src="https://1.bp.blogspot.com/-qu2HUszllsc/YS56k_XyrNI/AAAAAAAAK_k/MHazLyq5P-EY4cmcrfJtZBAxKjOCHSd4wCLcBGAsYHQ/w400-h220/shutterstock_1207894222%2B%25283%2529.jpg" width="400" /></a></div><div><i><span style="color: #333333; font-family: "inherit",serif; mso-bidi-font-family: "Times New Roman"; mso-fareast-font-family: "Times New Roman";"><br /></span></i></div><div><i><span style="color: #333333; font-family: "inherit",serif; mso-bidi-font-family: "Times New Roman"; mso-fareast-font-family: "Times New Roman";">John Halamka, M.D., president, Mayo Clinic Platform, and Paul
Cerrato, senior research analyst and communications specialist, Mayo Clinic
Platform, wrote this article.</span></i></div>
<p class="MsoNormal" style="line-height: 150%;"><span style="color: #333333;">The promise of personalized medicine will eventually allow
clinicians to offer individual patients more precise advice on prevention,
early detection and treatment. Of course, the operative word is </span><i style="color: #333333;">eventually. </i><span style="color: #333333;">A
closer examination of the screening tools available to detect breast cancer
demonstrates that we still have a way to go before we can fulfill that promise.
But with the help of better technology, we are getting closer to that
realization.</span></p>
<p class="MsoNormal" style="line-height: 150%;"><span style="color: #333333;">Disease screening is about risk assessment. Researchers collect
data on thousands of patients who develop breast cancer, for instance, and
discover that the age range, family history and menstruation history of those
who develop the disease differs significantly from those who remain free of it.
That in turn allows policy makers to create a screening protocol that suggests
women of a certain age who have experienced early menarche or late menopause
are more likely to develop the malignancy. That risk assessment is consistent
with the fact that more reproductive years means more exposure to the hormones
that contribute to breast cancer. Similarly, there’s evidence to show that
women with first degree relatives with the cancer and those with a history of
ovarian cancer or HRT use are at greater risk.</span></p>
<p class="MsoNormal" style="line-height: 150%;"><span style="color: #333333;">Statistics like this are the basis for several breast cancer
risk scoring systems, including the Gail score, the IBIS score, and BCSC
tool. The </span><a href="https://bcrisktool.cancer.gov/"><span style="font-family: "inherit",serif; mso-bidi-font-family: "Times New Roman"; mso-fareast-font-family: "Times New Roman";">National
Cancer Institute</span></a><span style="color: #333333;">, which
uses the Gail model, explains: “The Breast Cancer Risk Assessment Tool allows
health professionals to estimate a woman's risk of developing invasive breast
cancer over the next 5 years and up to age 90 (lifetime risk). The tool uses a
woman’s personal medical and reproductive history and the history of breast
cancer among her first-degree relatives (mother, sisters, daughters) to
estimate absolute breast cancer risk—her chance or probability of developing
invasive breast cancer in a defined age interval.” While the screening tool
saves lives, it can also be misleading. If, for example, it finds that a woman
has a 1% likelihood of developing breast cancer, what that really means is a
large population of women with those specific risk factors has a one in 100
risk of developing the disease. There is no way of knowing what the threat is
for any one patient in that group. Similar problems exist for the International
Breast Cancer Intervention Study </span><a href="https://ibis.ikonopedia.com/"><span style="font-family: "inherit",serif; mso-bidi-font-family: "Times New Roman"; mso-fareast-font-family: "Times New Roman";">(IBIS) score</span></a><span style="color: #333333;">, based on the
Tyrer-Cuzick Model, and the </span><a href="https://tools.bcsc-scc.org/BC5yearRisk/intro.htm"><span style="font-family: "inherit",serif; mso-bidi-font-family: "Times New Roman"; mso-fareast-font-family: "Times New Roman";">Breast Cancer Surveillance Consortium</span></a><span style="color: #333333;"> (BCSC) Risk Calculator. These
3 assessment tools can give patients a false sense of security if they don’t
dive into the details. BCSC, for instance, cannot be applied to women younger
that 35 or older than 74, nor does it accurately measure risk for anyone who
has previously had ductal carcinoma in situ (DCIS), or had breast augmentation.
Similarly, the NCI tool doesn’t accurately estimate risk in women with BRCA1 or
BRCA1 mutation, as well as certain other subgroups.</span></p>
<p class="MsoNormal"><span style="color: #333333;">During a conversation with Tufia Haddad, M.D,, a Mayo Clinic
medical oncologist with specialty interest in precision medicine in breast
cancer and artificial intelligence, she discussed the research she and her
colleagues are doing to improve the risk assessment process and identify more
high-risk women. Dr. Haddad pointed out that there are numerous obstacles that
prevent women from obtaining the best possible risk assessment. Too many women
do not have a primary care practitioner who might use a risk tool. And those
that do have a PCP are more likely to have an evaluation based on the Breast
Cancer Risk Assessment tool (the Gail model). “We prefer the Tyrer-Cuzick model
in part because it incorporates more personal information for each individual
patient including a detailed family history, a woman’s breast density from her
mammogram, as well as her history of atypia or other high risk benign breast
disease,” says Dr. Haddad. Unfortunately, the Tyrer-Cuzick method requires many
more data elements to assess breast cancer risk, which discourages busy
clinicians from using it.</span></p>
<p class="MsoNormal" style="line-height: 150%;"><span style="color: #333333;">Another obstacle to using any of these risk assessment tools is
the fact that they don’t readily fit into the average physician’s clinical
workflow. Ideally these tools should seamlessly integrate into the EHR system.
Even better would be the incorporation of AI-enhanced algorithms that automate
the abstraction of the required data elements from the patient’s record into
the assessment tool. For example, the algorithm would flag a family history of
breast cancer, increased breast density as determined during a mammogram, as
well as hormone replacement therapy and insert those risk factors into the Tyrer-Cuzick
tool.</span></p>
<p class="MsoNormal" style="line-height: 150%;"><span style="color: #333333;">Even with this AI-enhanced approach, all of the available risk
models fall short because they take a population-based approach, as we
mentioned above. Dr. Haddad and her colleagues are looking to make the
assessment process more individualized, as are others work in this specialty.
That model could incorporate each patient’s previous mammography results, their
genetics and benign breast biopsy findings, and much more. </span><a href="https://stm.sciencemag.org/content/13/578/eaba4373"><span style="font-family: "inherit",serif; mso-bidi-font-family: "Times New Roman"; mso-fareast-font-family: "Times New Roman";">Adam Yala,</span></a><span style="color: #333333;"> and his colleagues at
MIT recently developed a mammography-based deep learning model designed to take
this more sophisticated approach. Called Mirai, it was trained on a large data
set from Massachusetts General Hospital and from facilities in Sweden and
Taiwan. The new model generated
significantly better results for breast cancer risk prediction than the TC
model.</span></p>
<p class="MsoNormal"><span style="color: #333333;">Breast cancer risk assessment continues to evolve. And with
better utilization of existing assessment tools and the assistance of deep
learning, we can look forward to better patient outcomes.</span></p>John Halamkahttp://www.blogger.com/profile/04550236129132159307noreply@blogger.com0tag:blogger.com,1999:blog-4384692836709903146.post-25824238971067122322021-08-23T13:09:00.000-07:002021-08-23T13:09:21.535-07:00Can Social Determinants of Health Predict Your Patient’s Future?<h4 style="text-align: left;"><span style="font-family: inherit;">The evidence is mixed but suggests
that these overlooked variables have a profound impact on each patient’s
journey. </span></h4><div><div class="separator" style="clear: both; text-align: center;"><a href="https://1.bp.blogspot.com/-IkEfLfJC8pA/YR7oOqLA7uI/AAAAAAAAK4c/cRC8dGyfn0khiNAKV_FbTL_-Jt7WV9OIgCLcBGAsYHQ/s800/shutterstock_627703277.jpg" imageanchor="1" style="margin-left: 1em; margin-right: 1em;"><img border="0" data-original-height="533" data-original-width="800" height="266" src="https://1.bp.blogspot.com/-IkEfLfJC8pA/YR7oOqLA7uI/AAAAAAAAK4c/cRC8dGyfn0khiNAKV_FbTL_-Jt7WV9OIgCLcBGAsYHQ/w400-h266/shutterstock_627703277.jpg" width="400" /></a></div></div><p class="MsoNormal" style="line-height: 150%; margin-bottom: 0in;"><i><span style="line-height: 150%;"><span style="font-family: inherit;">This article was written by Tim
Suther, Nicole Hobbs, Jeff McGinn, Matt Turner with Change Healthcare, John
Halamka, MD, MS, president of Mayo Clinic Platform, and Paul Cerrato, senior
research analyst and communications specialist, Mayo Clinic Platform.</span></span></i></p>
<p class="MsoNormal" style="line-height: 150%; margin-bottom: 0in;"><span style="font-family: inherit; line-height: 150%;">By one estimate, social determinants
of health (SDoH) influence up to </span><a href="https://bmcmedinformdecismak.biomedcentral.com/articles/10.1186/1472-6947-14-36" style="font-family: inherit;"><span style="line-height: 150%;">80% of health outcomes</span></a><span style="font-family: inherit; line-height: 150%;">. Although reports like this suggest that these
social factors have a major impact, thought leaders continue to debate whether
they can also enhance the accuracy in predictive models. Resolving that debate
is far from simple because the answer depends on the type, source and quality
of the data, and the design of the model under consideration.</span></p>
<p class="MsoNormal" style="line-height: 150%; margin-bottom: 0in;"><span style="font-family: inherit;">In general, we derive SDoH from subjective
and objective sources. Subjective data includes self-reported or clinician-collected
data such as patient reported outcomes, Z codes from ICD-10-CM that report
factors that influence health status and interactions with health service
providers, and other unstructured EHR data. Objective data includes individual-level
and community-level data from government, public and private (and consumer
behavior) sources; it’s usually more structured and often derived from national-level
datasets.</span></p>
<p class="MsoNormal" style="line-height: 150%; margin-bottom: 0in;"><span style="font-family: inherit; line-height: 150%;">Unfortunately, the </span><a href="https://www.hmpgloballearningnetwork.com/site/frmc/article/using-predictive-analytics-address-social-determinants-health" style="font-family: inherit;"><span style="line-height: 150%;">research on the value of SDoH</span></a><span style="font-family: inherit; line-height: 150%;"> in predictive models varies widely. Some
studies report no appreciable differences when SDoH are injected into models,
while others report significant enhancements to predictive power.
Unsurprisingly, these varying study results depend in part on levels of reliance
on traditional clinical models and, most importantly, on the types and sources
of SDoH data employed in the studies.</span></p>
<p class="MsoNormal" style="line-height: 150%; margin-bottom: 0in;"><span style="font-family: inherit; line-height: 150%;">For example, a group from Johns
Hopkins Bloomberg School of Public Health demonstrated SDoH predictive models
can fail in part due to predictive model design as well as to EHR-level data
that is unstructured and collected inconsistently. They also demonstrated that dependence on data
from EHR-derived population health databases for SDoH can be problematic
because the data tends to be used as a proxy for individual-level social factors.
The problem lies in the fact that these
proxies are often based on assumptions, not evidence. </span><a href="https://healthitanalytics.com/news/top-3-data-challenges-to-addressing-the-social-determinants-of-health" style="font-family: inherit;"><span style="line-height: 150%;">Other research</span></a><span style="font-family: inherit; line-height: 150%;"> supports the above and showcases the
challenges of using SDoH data from sources that traditionally struggle with the
comprehensive collection and standardization of these data types.</span></p>
<p class="MsoNormal" style="line-height: 150%; margin-bottom: 0in;"><span style="font-family: inherit; line-height: 150%;">On a more positive note, </span><a href="https://bmcpublichealth.biomedcentral.com/articles/10.1186/s12889-020-08735-0" style="font-family: inherit;"><span style="line-height: 150%;">several studies</span></a><span style="font-family: inherit; line-height: 150%;"> </span><a href="https://pubmed.ncbi.nlm.nih.gov/21499815/" style="font-family: inherit;"><span style="line-height: 150%;">and</span></a><span style="font-family: inherit; line-height: 150%;">
</span><a href="https://healthitanalytics.com/news/how-social-determinants-data-can-enhance-machine-learning-tools" style="font-family: inherit;"><span style="line-height: 150%;">healthcare articles</span></a><span style="font-family: inherit; line-height: 150%;"> have reported success by relying on objectively
collected and/or highly structured and consistent data. For example, </span><a href="https://pubmed.ncbi.nlm.nih.gov/21499815/" style="font-family: inherit;"><span style="line-height: 150%;">one study</span></a><span style="font-family: inherit; line-height: 150%;"> that used EHR-derived SDoH data sources found that the addition of
structured data on median income, unemployment rate, and education from trustworthy non-EHR sources
enhanced their model’s health prediction
granularity <span style="background: white;">for
some of the most vulnerable subgroups of patients.</span> </span><a href="https://pubmed.ncbi.nlm.nih.gov/31985729/" style="font-family: inherit;"><span style="line-height: 150%;">In another study</span></a><span style="font-family: inherit; line-height: 150%;">, collaboration between Stanford, Harvard, and the Imperial
College London found that adding structured SDoH data from the US Census, along
with using machine learning techniques, improved risk prediction model accuracy
for hospitalization, death, and costs. They also showed that their models based
on SDoH alone, as well as those based on clinical comorbidities alone, could predict
health outcomes and costs. Similarly, </span><a href="https://bmcmedinformdecismak.biomedcentral.com/articles/10.1186/1472-6947-14-36" style="font-family: inherit;"><span style="line-height: 150%;">researchers at The Ohio State University
College</span></a><span style="font-family: inherit; line-height: 150%;"> of Medicine
added community-level and consumer behavior data not available in standard EHR
data and found it enhanced the study of and impact on obesity prevention. </span><a href="https://pubmed.ncbi.nlm.nih.gov/21499815/" style="font-family: inherit;"><span style="line-height: 150%;">Juhn et. al. at Mayo Clinic</span></a><span style="font-family: inherit; line-height: 150%;"> tapped telephone survey data and appended housing and
neighborhood characteristic data from local government sources to create a
socioeconomic status index (HOUSES). They first showed that HOUSES correlated
well with outcome measures and </span><a href="https://pubmed.ncbi.nlm.nih.gov/31985729/" style="font-family: inherit;"><span style="line-height: 150%;">later showed that HOUSES</span></a><span style="font-family: inherit; line-height: 150%;"> could even serve as a predictive tool for graft
failure in patients.</span></p>
<p class="MsoNormal" style="line-height: 150%; margin-bottom: 0in;"><b><span style="line-height: 150%;"><span style="font-family: inherit;">Patient Level SDoH + Clinical Data =
Predictive Power</span></span></b></p>
<p class="MsoNormal" style="line-height: 150%; margin-bottom: 0in;"><span style="font-family: inherit;">Incorporating social factors into the healthcare equation can
fill gaps needed at the point of care, but it also generates better healthcare predictions,
but only when these determinants are patient level and linked to robust clinical
data. Change Healthcare, for example, has curated such an integrated
national-level dataset, linking billions of historical de-identified distinct
medical claims with patient-level social, physical and behavioral determinants
of health. One of this dataset’s most important uses is to understand the
relative weight of specific patient SDOH factors, in comparison to clinical
factors alone, for various therapeutic conditions, including COVID-19. For
example, across Change Healthcare’s research, economic stability is repeatedly
ranked as the highest or among the highest predictors of the healthcare
experience. Despite this realization, most end users, including providers and
payers, lack such visibility (or rely on geographic averages that are unhelpful
in making accurate predictive models).</span></p>
<p class="MsoNormal" style="line-height: 150%; margin-bottom: 0in;"><span style="line-height: 150%;"><span style="font-family: inherit;">Incorporating SDoH data into
predictive models holds much promise. Given the relative newness of SDoH data
in predictive analytics, along with a lack of data standardization and scale, it’s
not surprising to find varying degrees of success in using it to improve predictive
health models. But as researchers learn more about the best types and sources of
SDoH data to use, along with developing better-suited models for these types of
data, we’re likely to see significant advances in healthcare predictive models.
By combining the right data with the right models, SDoH are a powerful asset in
predictive models of health, outcomes, and potential health disparities.</span></span></p><p class="MsoNormal" style="line-height: 150%; margin-bottom: 0in;"><b>If you're still with us . . .</b></p><p class="MsoNormal" style="line-height: 150%; margin-bottom: 0in;">Please consider supporting <span style="font-family: inherit;"><a href="https://www.linkedin.com/in/steve-parodi-95263622">Dr. Steve Parodi</a>, <a href="https://www.linkedin.com/in/reedabelson/">Reed Abelson</a> and I</span> by "voting up" on our panel at the upcoming <a href="https://www.sxsw.com/conference/">South by Southwest conference</a> in March of 2022. Our proposed panel, "Extending the Stethoscope Into the Home," will<span style="font-family: inherit;"> dive into a discussion about acute health
care for patients in their home and the infrastructures needed to support it. If you are so inclined to vote, please do so <a href="https://panelpicker.sxsw.com/vote/117442">here</a>.</span></p>John Halamkahttp://www.blogger.com/profile/04550236129132159307noreply@blogger.com0tag:blogger.com,1999:blog-4384692836709903146.post-46936967425805109402021-08-17T13:54:00.000-07:002021-08-17T13:54:49.739-07:00We Need to Open Up the AI Black Box<h4 style="text-align: left;">To convince physicians and nurses that deep
learning algorithms are worth using in everyday practice, developers need to
explain how they work in plain clinical English.</h4><div class="separator" style="clear: both; text-align: center;"><a href="https://1.bp.blogspot.com/-6I6I3wRPX7g/YRrxERtRwLI/AAAAAAAAK3M/4HCuBIAkv3kZZzps0Vf-IOKX5uJ0inBEgCLcBGAsYHQ/s1200/Do-We-Need-to-Open-the-AI-Black-Box.jpg" imageanchor="1" style="margin-left: 1em; margin-right: 1em;"><img border="0" data-original-height="800" data-original-width="1200" height="266" src="https://1.bp.blogspot.com/-6I6I3wRPX7g/YRrxERtRwLI/AAAAAAAAK3M/4HCuBIAkv3kZZzps0Vf-IOKX5uJ0inBEgCLcBGAsYHQ/w400-h266/Do-We-Need-to-Open-the-AI-Black-Box.jpg" width="400" /></a></div><div><p class="MsoNormal" style="line-height: 150%; mso-margin-bottom-alt: auto; mso-margin-top-alt: auto;"><i><span style="color: #333333; mso-bidi-font-family: Calibri; mso-bidi-theme-font: minor-latin; mso-fareast-font-family: "Times New Roman";">Paul
Cerrato, senior research analyst and communications specialist, Mayo Clinic
Platform, and John Halamka, M.D., president, Mayo Clinic Platform, wrote this
article.<o:p></o:p></span></i></p>
<p class="MsoNormal" style="line-height: 150%; mso-margin-bottom-alt: auto; mso-margin-top-alt: auto;"><span style="color: #333333;">AI’s so-called black box refers
to the fact that much of the underlying technology behind machine learning-enhanced
algorithms is </span><span style="border: 1pt none windowtext; padding: 0in;">probability/statistics
without a human readable explanation. </span><span style="color: #333333;">Oftentimes that’s the case because the advanced math or the data
science behind the algorithms is too complex for the average user to understand
without additional training. Several stakeholders in digital health maintain, however,
that this lack of understanding isn’t that important. They argue that as long
as an algorithm generates actionable insights, most clinicians don’t really
care about what’s “under the hood.” Is that reasoning sound?</span></p>
<p class="MsoNormal" style="line-height: 150%; mso-layout-grid-align: none; text-autospace: none;"><span style="color: #333333; mso-bidi-font-family: Calibri; mso-bidi-theme-font: minor-latin; mso-fareast-font-family: "Times New Roman";">Some
thought leaders point to the fact that there are many advanced,
computer-enhanced diagnostic and therapeutic tools currently in use that
physicians don’t fully understand, but nonetheless accept. <span style="mso-spacerun: yes;"> </span>The CHA2DSA-VASc score, for instance, is used
to estimate the likelihood of a patient with non-valvular atrial fibrillation
having a stroke. Few clinicians are familiar with the original research or detailed
reasoning upon which the calculator is based, but they nonetheless use the
tool. <span style="mso-spacerun: yes;"> </span>Similarly, many physicians use the
FRAX score to estimate a patient’s 10-year risk of developing a bone fracture,
despite the fact that they have not investigated the underlying math.<span style="mso-spacerun: yes;"> </span><o:p></o:p></span></p>
<p class="MsoNormal" style="line-height: 150%; mso-layout-grid-align: none; text-autospace: none;"><span style="color: #333333;">It’s
important to point out, however, that the stroke risk tool and the FRAX tool
both have major endorsements from organizations that physicians respect. The </span><a href="https://www.acc.org/latest-in-cardiology/articles/2014/10/14/11/02/the-revised-acc-aha-hrs-guidelines-for-the-management-of-patients-with-atrial-fibrillation">American Heart Association</a><span style="color: #333333;"> and the American College of
Cardiology both recommend the CHA2DSA-VASc score while the National
Osteoporosis Foundation supports the use of FRAX score. That gives physicians
confidence in these tools even if they don’t grasp the underlying details. To
date, there are no major professional associations recommending specific AI-enabled
algorithms to supplement the diagnosis or treatment of disease. The </span><a href="https://pubmed.ncbi.nlm.nih.gov/31862754/">American
Diabetes Association</a><span style="color: #333333;"> did
include a passing mention of an AI-based screening tool in its 2020 Standards
of Medical Care in Diabetes, stating: “</span>Artificial intelligence systems that
detect more than mild diabetic retinopathy and diabetic macular edema
authorized for use by the FDA represent an alternative to traditional screening
approaches. However, the benefits and optimal utilization of this type of
screening have yet to be fully determined.” That can hardly be considered a
recommendation.</p>
<p class="MsoNormal" style="line-height: 150%; mso-layout-grid-align: none; text-autospace: none;"><span style="color: #333333;">Given
this scenario, most physicians have reason to be skeptical, and surveys bear
out that skepticism. A </span><a href="https://academic.oup.com/jamia/article-abstract/28/6/1117/6046152?redirectedFrom=fulltext">survey of 91 primary care physicians</a><span style="color: #333333;"> found that <i>understandability
</i>of AI is one of the important attributes they want before trusting its
recommendations during breast cancer screening. Similarly, a survey of senior
specialists in UK found that understandability was one of their primary
concerns about AI. </span><a href="https://pubmed.ncbi.nlm.nih.gov/32106285/">Among
New Zealand physicians</a><span style="color: #333333;">,
88% </span>were
more likely to trust an AI algorithm that produced an understandable
explanation of its decisions.</p><p class="MsoNormal" style="line-height: 150%; mso-layout-grid-align: none; text-autospace: none;"><span style="color: #333333;">Of
course, it may not be possible to fully explain the advanced mathematics used
to create machine learning based algorithms. But there are other ways to describe
the logic behind these tools that would satisfy clinicians. As we have mentioned in previous publications
and oral presentations, there are tutorials available to simplify machine
learning-related systems like neural networks, random forest modeling,
clustering, and gradient boosting. Our </span><a href="https://www.taylorfrancis.com/books/mono/10.1201/9781003094234/digital-reconstruction-healthcare-paul-cerrato-john-halamka">most recent book</a><span style="color: #333333;"> contains an entire chapter on this
digital toolbox. Similarly, <i>JAMA </i> has created </span><a href="https://sites.jamanetwork.com/machine-learning/#multimedia">clinician friendly video tutorials</a><span style="color: #333333;"> designed to graphically
illustrate how deep learning is used in medical image analysis and how such
algorithms can be used to help detect lymph node metastases in breast cancer
patients. </span></p>
<p class="MsoNormal" style="line-height: 150%;"><span style="color: #333333; mso-bidi-font-family: Calibri; mso-bidi-theme-font: minor-latin; mso-fareast-font-family: "Times New Roman";">These resources require clinicians to take the initiative and
learn a few basic AI concepts, but developers and vendors also have an
obligation to make their products more transparent.<span style="mso-spacerun: yes;"> </span>One way to accomplish that goal is through </span><a href="https://www.nature.com/articles/s42256-021-00338-7"><span style="mso-bidi-font-family: Calibri; mso-bidi-theme-font: minor-latin; mso-fareast-font-family: "Times New Roman";">saliency maps and generative adversarial
networks</span></a><span style="color: #333333; mso-bidi-font-family: Calibri; mso-bidi-theme-font: minor-latin; mso-fareast-font-family: "Times New Roman";">. Using
such techniques, it’s possible to highlight the specific pixel grouping that a neural
network has identified as a trouble spot, which the clinician can then view on
a radiograph, for example.<span style="mso-spacerun: yes;"> </span>Alex DeGrave,
with the University of Washington, and his colleagues, used this approach to
help explain why an algorithm designed to detect COVID-19-related changes in
chest X-rays made its recommendations. </span><a href="https://www.nature.com/articles/s41746-019-0216-8"><span style="mso-bidi-font-family: Calibri; mso-bidi-theme-font: minor-latin; mso-fareast-font-family: "Times New Roman";">Amirata Ghrobani and associates</span></a><span style="color: #333333; mso-bidi-font-family: Calibri; mso-bidi-theme-font: minor-latin; mso-fareast-font-family: "Times New Roman";"> from Stanford University have
taken a similar approach to help clinicians comprehend the echocardiography
recommendations coming from a deep learning system. </span><span style="mso-bidi-font-family: Calibri; mso-bidi-theme-font: minor-latin;">The
researchers trained a convolutional neural network (CNN) on over 2.6 million echocardiogram
images from more than 2,800 patients and demonstrated it was capable of
identifying enlarged left atria, left ventricular hypertrophy, and several
other abnormalities. To open up the black box, Ghorbani et al presented readers
with “biologically plausible regions of interest” in the echocardiograms they
analyzed so they could see for themselves the reason for the interpretation
that the model has arrived at. For instance, if the CNN said it had identified
a structure such as a pacemaker lead, it highlighted the pixels it identifies
as the lead. Similar clinician-friendly images are presented for a severely
dilated left atrium and for left ventricular hypertrophy. <o:p></o:p></span></p>
<p class="MsoNormal" style="line-height: 150%; mso-margin-bottom-alt: auto; mso-margin-top-alt: auto;"><span style="color: #333333; mso-bidi-font-family: Calibri; mso-bidi-theme-font: minor-latin; mso-fareast-font-family: "Times New Roman";">Deep
learning systems are slowly ushering in a new way to manage diagnosis and treatment,
but to bring skeptical clinicians on board, we need to pull the curtain back. In
addition to providing evidence that these tools are equitable and clinically
effectively, practitioners want reasonable explanations to demonstrate that they
will do what they claim to do.</span></p></div>John Halamkahttp://www.blogger.com/profile/04550236129132159307noreply@blogger.com0tag:blogger.com,1999:blog-4384692836709903146.post-4183786464857327212021-07-30T10:09:00.000-07:002021-07-30T10:09:34.213-07:00The Future Belongs to Digital Pathology<h4 style="text-align: left;"><span style="font-family: inherit;">Advances in artificial intelligence
are slowly transforming the specialty, much the way radiology is being
transformed by similar advances in digital technology.</span></h4><div><div class="separator" style="clear: both; text-align: center;"><a href="https://1.bp.blogspot.com/-NnmIqwNsJbk/YQQx-7rajFI/AAAAAAAAKvo/1NeMoO7Kua8SgIYjNylUY4AYBbgXWpoXwCLcBGAsYHQ/s800/Pathology.jpg" imageanchor="1" style="margin-left: 1em; margin-right: 1em;"><img border="0" data-original-height="532" data-original-width="800" height="266" src="https://1.bp.blogspot.com/-NnmIqwNsJbk/YQQx-7rajFI/AAAAAAAAKvo/1NeMoO7Kua8SgIYjNylUY4AYBbgXWpoXwCLcBGAsYHQ/w400-h266/Pathology.jpg" width="400" /></a></div><i><span style="font-family: inherit;"><br />John Halamka, M.D., president, Mayo Clinic Platform, and Paul
Cerrato, senior research analyst and communications specialist, Mayo Clinic
Platform, wrote this article.</span></i></div>
<p class="MsoNormal" style="line-height: 150%;"><span style="font-family: inherit;">Any patient who faces a potential
cancer diagnosis knows how important an accurate, timely pathology report is. Similarly, surgeons often require fast
pathology results when they are performing a delicate procedure to determine
their course of action during an operation. New technological developments are
poised to meet the needs of patients and clinicians alike.</span></p>
<p class="MsoNormal" style="line-height: 150%;"><span style="font-family: inherit;">AI can improve pathology practice in
numerous ways. The right digital tools can automate several repetitive tasks,
including the detection of small foci. It can also help improve the staging of
many malignancies, make the workflow process more efficient, and help classify
images, which in turn gives pathologists a “second set of eyes”. And those
“eyes” do not grow tired at the end of a long day or feel stressed out from too
much work.</span></p>
<p class="MsoNormal" style="line-height: 150%;"><span style="font-family: inherit;">Such capabilities have far-reaching
implications. With the right scanning hardware and the proper viewer software,
pathologists and technicians can easily view and store whole slide images
(WSIs). That view is markedly different from what they see through a
microscope, which only allows a narrow field of view. In addition, digitization
allows pathologists to mark up WSIs with non-destructive annotations, use the
slides as teaching tools, search a laboratory’s archives to make comparisons
with images that depict similar cases, give colleagues and patients access to
the images, and create predictive models. And if the facility has cloud storage
capabilities, it allows clinicians, patients, and pathologists around the world
to access the data.</span></p>
<p class="MsoNormal" style="line-height: 150%;"><span style="font-family: inherit;">A 2020 prospective trial conducted by University of Michigan and
Columbia University investigators illustrates just how profound the impact of
AI and ML can be when applied to pathology. Todd Hollon and colleagues point out that
interoperative diagnosis of cancer relies on a “contracting, unevenly
distributed pathology workforce.”<sup>1</sup> The process can be quite
inefficient, requiring a tissue specimen travel from the OR to a lab, followed
by specimen processing, slide preparation by a technician, and a pathologist’s
review. At University of Michigan, they are now using Stimulated Raman
histology, an advanced optical imaging method, along with a convolutional
neural network (CNN) to help interpret the images. The machine learning tools were
trained to detect 13 histologic categories and includes an inference algorithm
to help make a diagnosis of brain cancer. Hollon et al conducted a 2-arm,
prospective multicenter, non-inferiority trial to compare the CNN results to
those of human pathologists. The trial, which evaluated 278 specimens,
demonstrated that the machine learning system was as accurate as pathologists’
interpretation (94.6% vs 93.9%). Equally important was the fact that it took
under 15 seconds for surgeons to get their results with the AI system, compared
to 20-30 minutes with conventional techniques. And that latter estimate does
not represent the national average. In some community settings, slides have to
be shipped by special courier to labs that are hours away.</span></p>
<p class="MsoNormal" style="line-height: 150%;"><span style="font-family: inherit;">Mayo Clinic is among several
forward-thinking health systems that are in the process of implementing a
variety of digital pathology services. <span style="color: #201f1e;">Mayo Clinic has partnered with Google and is leveraging their
technology in two ways. The program will extend Mayo Clinic’s comprehensive
Longitudinal Patient Record profile with digitized pathology images to better
serve and care for patients. And we are exploring new search capabilities to
improve digital pathology analytics and AI. The Mayo/Google project is being
conducted with the help of Sectra, a digital slide review and image storage and
management system. </span><span style="background: white; color: #202426;">Once proof of concept, system testing, and configuration
activities are complete, the digital pathology solution will be introduced
gradually to Mayo Clinic departments throughout Rochester, Florida, and
Arizona, as well as the Mayo Clinic Health System.</span></span></p>
<p class="MsoNormal" style="line-height: 150%;"><span style="font-family: inherit;">The new digital capabilities taking
hold in several pathology labs around the globe are likely to solve several
vexing problems facing the specialty. Currently there is a <a href="https://clpmag.com/diagnostic-technologies/digital-pathology/digital-pathology-gives-rise-computational-pathology/">shortage
of pathologists worldwide</a>, and in some countries, that shortage is severe.
One estimate found there is one pathologist per 1.5 million people in parts of
Africa. And China has one fourth the number of pathologists practicing in the
U.S., on a per capita basis. Studies predict that the steady decline of the
number of pathologists in the U.S. will continue over the next two decades. A
lack of subspecialists is likewise a problem. Similarly, there are reports of
poor accuracy and reproducibility, with many practitioners making subjective
judgements based on a manual estimate of the percentage of positive cells for a
biomarker. Finally, there is reason to believe that implementing digital
pathology systems will likely improve a health system’s financial return on
investment. One study has suggested that it can <span style="background: white; color: #424242;">“improve the efficiency of pathology workloads
by 13%.”</span><span style="background: white; color: #424242; line-height: 150%;"> <sup>2</sup></span></span></p>
<p class="MsoNormal" style="line-height: 150%;"><span style="font-family: inherit;">As we have said several times in these
columns, AI and ML are certainly not a panacea, and they will never replace an
experienced clinician or pathologist. But taking advantage of the tools
generated by AI/ML will have a profound impact of diagnosis and treatment for
the next several decades.</span></p>
<p class="MsoNormal" style="line-height: 150%;"><span style="mso-bidi-font-family: Calibri; mso-bidi-theme-font: minor-latin;"><o:p><span style="font-family: inherit;"> </span></o:p></span></p>
<p class="MsoNormal" style="line-height: 150%;"><b><span style="font-family: inherit;">References</span></b></p><p class="MsoNormal" style="line-height: 150%; text-indent: 0px;"><span style="font-family: inherit;"><span style="text-indent: -0.5in;">1. </span><span style="text-indent: -0.5in;">Hollon T, Pandian B, Adapa A et al. Near real-time intraoperative
brain tumor diagnosis using stimulated Raman histology and deep neural
networks. </span><i style="text-indent: -0.5in;">Nat. Med. </i><span style="text-indent: -0.5in;">2020. 26:52-58.</span></span></p><p class="MsoNormal" style="line-height: 150%; text-indent: 0px;"><span style="font-family: inherit;"><span style="text-indent: -0.5in;">2. </span><span style="color: #424242; text-indent: -0.5in;">Ho J, Ahlers SM, Stratman
C, et al. Can digital pathology result in cost savings? a financial projection
for digital pathology implementation at a large integrated health care
organization. </span><i style="color: #424242; text-indent: -0.5in;"><span style="border: none windowtext 1.0pt; mso-border-alt: none windowtext 0in; padding: 0in;">J Pathol Inform</span></i><span style="color: #424242; text-indent: -0.5in;">. 2014;5(1):33; doi:
10.4103/2153-3539.139714.</span></span></p>John Halamkahttp://www.blogger.com/profile/04550236129132159307noreply@blogger.com0tag:blogger.com,1999:blog-4384692836709903146.post-36493917459792528492021-07-28T09:14:00.005-07:002021-07-28T09:14:43.801-07:00Shift Happens<h4 style="text-align: left;">Dataset shift can thwart the best
intentions of algorithm developers and tech-savvy clinicians, but there are
solutions.</h4><div class="separator" style="clear: both; text-align: center;"><a href="https://1.bp.blogspot.com/-s-fay2XyN68/YQGCFp9FzII/AAAAAAAAKuc/L8H5qu5SM8gASZQLZ3eiKJbK-4kiVjnIACPcBGAYYCw/s800/Shift-Happens.jpg" imageanchor="1" style="margin-left: 1em; margin-right: 1em;"><img border="0" data-original-height="533" data-original-width="800" height="266" src="https://1.bp.blogspot.com/-s-fay2XyN68/YQGCFp9FzII/AAAAAAAAKuc/L8H5qu5SM8gASZQLZ3eiKJbK-4kiVjnIACPcBGAYYCw/w400-h266/Shift-Happens.jpg" width="400" /></a></div><p class="MsoNormal" style="line-height: 150%; text-align: left;"><o:p></o:p></p>
<p class="MsoNormal" style="line-height: 150%; text-align: left;"><i><span style="font-family: Georgia, serif; font-size: 10pt; line-height: 150%;">John Halamka, M.D.,
president, Mayo Clinic Platform, and Paul Cerrato, senior research analyst and communications
specialist, Mayo Clinic Platform, wrote this article.</span></i></p><p class="MsoNormal">Generalizability has always been a concern in health care,
whether we’re discussing the application of clinical trials or machine-learning
based algorithms. A large randomized controlled trial that finds an intensive
lifestyle program doesn’t reduce the risk of cardiovascular complications in
Type 2 diabetics, for instance, suggests the diet/exercise regimen is not worth
recommending to patients. But the question immediately comes to mind: Can that
finding be generalized to the entire population of Type 2 patients? As we have
pointed out in <a href="https://www.routledge.com/Reinventing-Clinical-Decision-Support-Data-Analytics-Artificial-Intelligence/Cerrato-Halamka/p/book/9781032081854">other publications</a>, subgroup analysis has demonstrated that many
patients do, in fact, benefit from such a program.<o:p></o:p></p><p class="MsoNormal">The same problem exists in health care IT. Several
algorithms have been developed to help classify diagnostic images, predict
disease complications, and more. A closer look at the datasets upon which these
digital tools are based indicates many suffer from dataset shift. In simple
English, dataset shift is what happens when the data collected during the
development of an algorithm changes over time and is different from the data
when the algorithm is eventually implemented. For example, the patient
demographics used to create a model may no longer represent the patient
population when the algorithm is put into clinical use. This happened when
COVID 19 changed the demographic characteristics of patients, making the Epic
sepsis prediction tool ineffective.<o:p></o:p></p><p class="MsoNormal"><a href="https://www.nejm.org/doi/full/10.1056/NEJMc2104626">Samuel Finlayson, PhD</a>, with Harvard Medical School, and his
colleagues described a long list of data set shift scenarios that can
compromise the accuracy and equity of AI-based algorithms, which in turn can
compromise patient outcomes and patient safety. Finlayson et al list 14
scenarios, which fall into 3 broad categories: changes in technology; changes
in population and setting; and changes in behavior. Examples of ways in which
dataset shift can create misleading outputs that send clinicians down the wrong
road include:<o:p></o:p></p><p class="MsoNormal"></p><ul style="text-align: left;"><li>Changes in the X-ray scanner models used</li><li>Changes in the way diagnostic codes are collected (e.g.
using ICD9 and then switching to ICD10)</li><li>Changes in patient population resulting from hospital
mergers</li></ul><o:p></o:p><p></p><p class="MsoNormal"><o:p></o:p></p><p class="MsoNormal"><o:p></o:p></p><p class="MsoNormal">Other potential problems to be cognizant of include changes
in your facility’s EHR system. Sometimes updates to the system may result in
changes in how terms are defined, which in turn can impact predictive algorithms
that rely on those definitions. If a term like elevated temperature or fever is
changed to pyrexia in one of the EHR drop down menus, for example, it may no
longer map to the algorithm that uses elevated temperature as one of the
variable definitions to predict sepsis, or any number of common infections.
Similarly, if the ML-based model has been trained on a patient dataset for a
medical specialty practice or hospital cohort, it’s likely that data will
generate misleading outputs when applied to a primary care setting.</p><p class="MsoNormal">Finlayson et al mention another example to be aware of:
changes in the way physicians practice can influence data collection: “Adoption
of new order sets, or changes in their timing, can heavily affect predictive
model output.” Clearly, problems like this necessitate strong interdisciplinary
ties, including an ongoing dialogue between the chief medical officer, clinical
department heads, and chief information officer and his or her team. Equally
important is the need for clinicians in the trenches to look for subtle changes
in practice patterns that can impact the predictive analytics tools currently
in place. Many dataset mismatches can be solved by updating variable mapping,
retraining or redesigning the algorithm, and multidisciplinary root cause
analysis.<o:p></o:p></p><p class="MsoNormal">While addressing dataset shift issues will improve the
effectiveness of your AI-based algorithms, they are only one of many stumbling
blocks to contend with. One classic example that demonstrates that computers
are still incapable to matching human intelligence is <a href="https://www.pulmonologyadvisor.com/home/topics/practice-management/the-potential-pitfalls-of-machine-learning-algorithms-in-medicine/">the study</a> that concluded
that patients with asthma are less likely to die from pneumonia that those who
don’t have asthma. The machine learning tool used to come to that unwarranted
conclusion had failed to take into account the fact that many asthmatics often
get faster, earlier, more intensive treatment when their condition flares up,
which results in a lower mortality rate. Had clinicians acted on the misleading
correlation between asthma and fewer deaths from pneumonia, they might have
decided asthma patients don’t necessarily need to hospitalized when they
develop pneumonia.<o:p></o:p></p><p class="MsoNormal">This kind of misdirection is relatively common and
emphasizes the fact that ML-enhanced tools sometimes have trouble separating
useless “noise” from meaningful signal. Another example worth noting: <a href="https://www.nature.com/articles/s42256-021-00338-7">Some algorithms</a> designed to help detect COVID 19 by analyzing X-rays suffer from
this shortcoming. Several of these deep learning algorithms <a href="https://www.itnonline.com/content/medical-ai-models-rely-shortcuts-could-lead-misdiagnosis-covid-19">rely on confounding variables instead of focusing on medical pathology</a>, giving clinicians the
impression that they are accurately identifying the infection or ruling out its
presence. Unbeknownst to their users, the algorithms have been shown to rely on
text markers or patient positioning instead of pathology findings.<o:p></o:p></p><p class="MsoNormal">At Mayo Clinic, we have had to address similar problems. A
palliative care model that was trained on data from the Rochester, Minnesota,
community, for instance, did not work well in our health system because the
severity of patient disease in a tertiary care facility is very different than
what’s seen in a local community hospital. Similarly, one of our algorithms
broke when a vendor did a point release in its software and changed the format
of the results. We also had a vendor with a CT stroke detection algorithm run
10 of our known stroke patients through its system and was only able to
identify one patient. The root cause: Mayo Clinic medical physicists have
optimized our radiation exposure to 25% of industry standards to reduce
radiation exposure to patients, but that changed the signal to noise ratio of
the images and the vendor’s system wasn’t trained on that ratio and couldn’t
find the images. <o:p></o:p></p><p class="MsoNormal"><a href="https://link.springer.com/epdf/10.1007/s00134-021-06473-4?sharing_token=BTnbFbW0KM5CYxPn1qaCdPe4RwlQNchNByi7wbcMAY4DrjdxYW4hKTo60CW2GATQvYAfEZww5c5HNOYCgver-68XLbgXA5n9H7bybR-Xdn54ljTJGmyt66xMWYp-WEV3p5K10nIM-roRWJkHakeE_xwOt_tpZYPRINQ0MmXPlvc%3D">Valentina Bellini, with University of Parma</a>, Parma, Italy,
and her colleagues sum up the AI shortcut dilemma in a graphic that illustrates
3 broad problem areas: Poor quality data, ethical and legal issues, and lack of
educational programs for clinicians who may be skeptical or uninformed about
the value and limitations of AI enhanced algorithms in intensive care settings.<o:p></o:p></p><p class="MsoNormal" style="line-height: 150%;">
</p><p class="MsoNormal">As we have pointed out in other blogs, ML-based algorithms
rely on math, not magic. But when reliance on that math overshadows clinicians’
diagnostic experience and common sense, they need to partner with their IT
colleagues to find ways to reconcile artificial and human intelligence.<o:p></o:p></p>
John Halamkahttp://www.blogger.com/profile/04550236129132159307noreply@blogger.com0tag:blogger.com,1999:blog-4384692836709903146.post-7452454714004652072021-07-23T08:42:00.000-07:002021-07-23T08:42:14.953-07:00Causality in Medicine: Moving Beyond Correlation in Clinical Practice<h4 style="line-height: 150%; text-align: left;">A
growing body of research suggests it’s time to abandon outdated ideas about how
to identify effective medical therapies.</h4>
<p class="MsoNormal" style="line-height: 150%;"><i><span style="color: #333333; mso-bidi-font-family: Calibri; mso-bidi-theme-font: minor-latin; mso-fareast-font-family: "Times New Roman";">Paul Cerrato, senior research analyst and communications
specialist, Mayo Clinic Platform, and John Halamka, M.D., president, Mayo
Clinic Platform, wrote this article.</span></i></p>
<p class="MsoNormal" style="line-height: 150%;">“Correlation
is not causation.” It’s a truism that researchers take for granted, and for
good reason. The fact that event A is followed by event B doesn’t mean that A
caused B. An observational study of 1,000 adults, for example, that found those
taking high doses of vitamin C were less likely to develop lung cancer doesn’t
prove the nutrient protects against the cancer; it’s always possible that a
third factor — a confounding variable — was responsible for both A and B. In other
words, patients taking lots of vitamin C may be less likely to get lung cancer
because they are more health conscious than the average person, and therefore more
likely to avoid smoking, which in turn reduces their risk of the cancer.</p><p class="MsoNormal" style="line-height: 150%;">As
this example illustrates, confounding variables are the possible contributing
factors that may mislead us into imagining a cause-and-effect relationship exists
when there isn’t one. It’s the reason interventional trials like the randomized
controlled trial (RCT) remain a more reliable way to determine causation than
observational studies. But it’s important to point out that in clinical
medicine, there are many treatment protocols in use that are not supported by
RCTs. Similarly, there are many risk factors associated with various diseases
but it’s often difficult to know for certain whether these risk factors are
actually contributing causes of said diseases. </p><p class="MsoNormal" style="line-height: 150%;">While
RCTs remain the good standard in medicine, they can be impractical for a
variety of reasons: they are often very expensive to perform; an RCT that exposes
patients to potentially harmful risk factor and compares them to those who
aren’t would be unethical; most trials require many exclusion and inclusion
criteria that don’t exist in the everyday practice of medicine. For instance,
they usually exclude patients with co-existing conditions, which may distort
the study results.</p><p class="MsoNormal" style="line-height: 150%;">One
way to address this problem is by accepting less than perfect evidence and using
a reliability scale or continuum to determine which treatments are worth using
and which are not. That scale might look something like this, with evidential
support growing stronger from left to right along the continuum: </p><div class="separator" style="clear: both; text-align: left;"><a href="https://1.bp.blogspot.com/-Q5CIdTEB7v0/YPcBnSx4I_I/AAAAAAAAKpw/M8E3VPFatQIia4eGkjThFGibWfRmFxDOACLcBGAsYHQ/s760/Capture.JPG" imageanchor="1" style="margin-left: 1em; margin-right: 1em;"><img border="0" data-original-height="209" data-original-width="760" height="110" src="https://1.bp.blogspot.com/-Q5CIdTEB7v0/YPcBnSx4I_I/AAAAAAAAKpw/M8E3VPFatQIia4eGkjThFGibWfRmFxDOACLcBGAsYHQ/w400-h110/Capture.JPG" width="400" /></a></div><div class="separator" style="clear: both; text-align: left;"><p class="MsoNormal" style="line-height: 150%;">In
the absence of RCTs, it’s feasible to consider using observational studies like
case/control and cohort trials to justify using a specific therapy. And while
such observational studies may still mislead because some confounding variables
have been overlooked, there are epidemiological criteria that strengthen the
weight given to these less than perfect studies:</p><p class="MsoNormal" style="line-height: 150%;"></p><ul style="text-align: left;"><li><span style="text-indent: -0.25in;">A stronger
association or correlation between two variables is more suggestive of a
cause/effect relationship than a weaker association.<br /><br /></span></li><li>Temporality. The
alleged effect must follow the suspected cause not the other way around. It
would make no sense to suggest that exposure to <i style="text-indent: -0.25in;"><span style="background: white; color: #202124; mso-bidi-font-family: Calibri; mso-bidi-theme-font: minor-latin; mso-fareast-font-family: "Times New Roman";">Mycobacterium
tuberculosis </span></i><span style="background: white; color: #202124; text-indent: -0.25in;">causes TB if all the cases of the infection occurred before
patients were exposed to the bacterium.<br /><br /></span></li><li>A dose-response relationship exists between alleged cause
and effect. For example, if researchers find that a blood lead level of
10 mcg/dl is associated with mild learning disabilities in children, 15 mcg/dl
is linked to moderate deficit, and 20 mcg/dl with severe deficits, this
gradient strengthens the argument for causality.<br /><br /></li><li>A biologically plausible mechanism of action
linking cause and effect strengthens the argument. In the case of lead
poisoning, there is evidence pointing to neurological damage brought on by
oxidative stress and a variety of other biochemical mechanisms.<br /><br /></li><li>Repeatability of the study findings: If the
results of one group of investigators are duplicated by independent
investigators, that lends further support to the cause/effect relationship.</li></ul><p class="MsoListParagraphCxSpLast" style="line-height: 150%; mso-layout-grid-align: none; mso-list: l0 level1 lfo1; text-autospace: none; text-indent: -.25in;"><o:p></o:p></p>
<p class="MsoNormal" style="line-height: 150%;">While
adherence to all these criteria suggests causality for observational studies, a
statistical approach called <a href="https://www.amazon.com/Book-Why-Science-Cause-Effect/dp/1541698967/ref=sr_1_1?dchild=1&gclid=CjwKCAjw55-HBhAHEiwARMCszuHEt7BrPpNAAf8yJDYRgpxZVG_2ji2jZ5Nr7Cf9WHqj6Yhf4zfGNBoCPz8QAvD_BwE&hvadid=241598182749&hvdev=c&hvlocphy=9004267&hvnetw=g&hvqmt=e&hvrand=1237753201018649440&hvtargid=kwd-814458155&hydadcr=22534_10353871&keywords=the+book+of+why&qid=1625846272&sr=8-1">causal
inference</a> can actually <i>establish </i>causality. The technique, which was spearheaded by <a href="https://samueli.ucla.edu/judea-pearl/">Judea Pearl, Ph.D.,</a> winner of the 2011 Turing Award, is considered
revolutionary by many thought leaders and will likely have profound
implications for clinical medicine, and for the role of AI and machine
learning. During the recent <a href="https://ce.mayo.edu/research/content/mayo-clinic-artificial-intelligence-symposium">Mayo
Clinic Artificial Intelligence Symposium</a>,
Adrian Keister, Ph.D., a senior data science analyst at Mayo Clinic, concluded
that causal inference is “possibly the most important advance in the scientific
method since the birth of modern statistics — maybe even more important than
that.”</p><p class="MsoNormal" style="line-height: 150%;"><o:p></o:p></p>
<p class="MsoNormal" style="line-height: 150%;">Conceptually,
causal inference starts with the conversion of word-based statements into
mathematical statements, with the help of a few new operators. While that may
sound daunting to anyone not well-versed in statistics, it’s not much different
than the way we communicate by using the language of arithmetic. A statement
like fifteen times five equals seventy five is converted to 15 x 5 = 75. In
this case, x is an operator. The new mathematical language of causal inference
might look like this if it were to represent an observational study that evaluated
the association between a new drug and an increase in patients’ lifespan: P
(L|D) where P is probability, L, lifespan, D is the drug, and | is an operator
that means “conditioned on.”</p>
<p class="MsoNormal" style="line-height: 150%;">An
interventional trial such as an RCT, on the other hand, would be written as X
causes Y if P (L|<i>do</i> (D)) > P(Y), in which case the <i>do</i>-operator
refers to the intervention, i.e., giving the drug in a controlled setting. This
formula is a way to of saying X (the drug being tested), causes Y (longer life)
if the results of the intervention are greater than the probability of a longer
life without administering the drug, in other words, the probability in the
placebo group, namely P(Y).</p>
<p class="MsoNormal" style="line-height: 150%;">This innovative
technique also uses causal graphs to show the relationship of a confounding
variable to a proposed cause/effect relationship. Using this kind of graph, one
can illustrate how the tool applies in a real-world scenario. Consider the
relationship between smoking and lung cancer. For decades, statisticians and
policy makers argued about whether smoking causes the cancer because all the
evidence supporting the link was observational. The graph would look something
like this.<br /><br /><o:p></o:p></p><p class="MsoNormal" style="line-height: 150%;"><b>Figure 1:</b></p></div><div class="separator" style="clear: both; text-align: left;"><a href="https://1.bp.blogspot.com/-ERt0XtAfZuM/YPcCWJ8JysI/AAAAAAAAKp4/24z9ttIV2iYu9tbvhgcr8i_tfrJDibLGACLcBGAsYHQ/s567/Blog-figure-2.JPG" imageanchor="1" style="margin-left: 1em; margin-right: 1em;"><img border="0" data-original-height="464" data-original-width="567" height="164" src="https://1.bp.blogspot.com/-ERt0XtAfZuM/YPcCWJ8JysI/AAAAAAAAKp4/24z9ttIV2iYu9tbvhgcr8i_tfrJDibLGACLcBGAsYHQ/w200-h164/Blog-figure-2.JPG" width="200" /></a></div><div class="separator" style="clear: both; text-align: left;">
<p class="MsoNormal" style="line-height: 150%;">G is
the confounding variable — a genetic predisposition for example — S is smoking and
LC is lung cancer. The implication here is that if a third factor causes persons
to smoke and causes cancer, one cannot necessarily conclude that smoking causes
lung cancer. What Pearl and his
associates discovered was that if an intermediate factor can be identified in
the pathway between smoking and cancer, it’s then possible to establish a cause/effect
relationship between the 2 with the help of a series of mathematical
calculations and a few algebraic rewrite tools. As figure 2 demonstrates, tar
deposits in the smokers’ lung are that intermediate factor. <br /><br /><o:p></o:p></p><p class="MsoNormal" style="line-height: 150%;"><b>Figure 2:</b></p></div><div class="separator" style="clear: both; text-align: left;"><a href="https://1.bp.blogspot.com/-0CjIHYURJZ0/YPcDLBPmmEI/AAAAAAAAKqI/4VNus2G6T2kQcZ2ktU5Jh8kwslNi3RZYgCLcBGAsYHQ/s1013/Blog-figure-3.JPG" imageanchor="1" style="margin-left: 1em; margin-right: 1em;"><img border="0" data-original-height="482" data-original-width="1013" src="https://1.bp.blogspot.com/-0CjIHYURJZ0/YPcDLBPmmEI/AAAAAAAAKqI/4VNus2G6T2kQcZ2ktU5Jh8kwslNi3RZYgCLcBGAsYHQ/s320/Blog-figure-3.JPG" width="320" /></a></div><br /><div class="separator" style="clear: both; text-align: left;">For a
better understanding of how causal inference works, Judea Pearl’s <a href="https://www.amazon.com/Book-Why-Science-Cause-Effect/dp/1541698967/ref=sr_1_1?dchild=1&gclid=CjwKCAjw55-HBhAHEiwARMCszuHEt7BrPpNAAf8yJDYRgpxZVG_2ji2jZ5Nr7Cf9WHqj6Yhf4zfGNBoCPz8QAvD_BwE&hvadid=241598182749&hvdev=c&hvlocphy=9004267&hvnetw=g&hvqmt=e&hvrand=1237753201018649440&hvtargid=kwd-814458155&hydadcr=22534_10353871&keywords=the+book+of+why&qid=1625846272&sr=8-1"><i>The Book
of Why</i></a><i> </i>is worth a closer look. It provides a plain English
explanation of causal inference. For a deeper dive, there’s <a href="https://www.amazon.com/Causal-Inference-Statistics-Judea-Pearl/dp/1119186846/ref=sr_1_3?crid=2V9DQDU9FQG4U&dchild=1&keywords=causal+inference+in+statistics+a+primer&qid=1625943484&sprefix=Causal+Inference%2Caps%2C209&sr=8-3"><i>Causal
Inference in Statistics: A Primer.</i></a></div><div class="separator" style="clear: both; text-align: left;">
<p class="MsoNormal" style="line-height: 150%;">Had
causal inference existed in the 1950s and 1960s, the argument by tobacco
industry lobbyists would have been refuted, which in turn might have saved many
millions of lives. The same approach holds tremendous potential as we begin to
apply it to predictive algorithms and other machine-learning based digital
tools. </p></div>John Halamkahttp://www.blogger.com/profile/04550236129132159307noreply@blogger.com0tag:blogger.com,1999:blog-4384692836709903146.post-25648121851496454392021-07-19T09:44:00.000-07:002021-07-19T09:44:00.805-07:00Taking Down the Fences that Divide Us<p><b>Innovation in healthcare requires new ways to
think about interdisciplinary solutions.</b></p>
<p class="MsoNormal" style="line-height: 150%;"><i>Paul Cerrato, senior research analyst and communications
specialist, Mayo Clinic Platform and </i><i>John Halamka, M.D., president, Mayo Clinic Platform, wrote
this article.</i></p>
<p class="MsoNormal" style="line-height: 150%;">During
the 10 years we have worked together, John and I have written often about the
power of words like <a href="https://www.amazon.com/Transformative-Power-Mobile-Medicine-Opportunities/dp/012814923X"><span style="color: #0563c1;">transformation</span></a>,
optimism, cynicism, and<a href="https://www.amazon.com/Reinventing-Clinical-Decision-Support-Intelligence/dp/0367186233/ref=cm_cr_arp_d_pdt_img_top?ie=UTF8"><span style="color: #0563c1;"> misdiagnosis</span></a>. Another
word that needs more attention is “interdisciplinary.” It’s been uttered so
many times in science, medicine, and technology that it’s lost much of its impact. We all give lip service to the idea, but many
aren’t willing or able to do the hard work required to make it a reality, and
one that fosters innovation and better patient care.</p>
<p class="MsoNormal" style="line-height: 150%;">Examples
of the dangers of focusing too narrowly on one discipline are all around us.
The disconnect between technology and medicine becomes obvious when you take a
closer look at the invention of blue light emitting diodes (LEDs), for
instance, for which Isamu Aksaki, Hiroshi Amano, and Shuti Nakamura won the
Nobel Prize in Physics in 2014. While this technology reinvented the way we
light own homes, providing a practical source of bright, energy-saving light,
the researchers failed to take into account the health effects of their
invention. Had they been encouraged to embrace
an interdisciplinary mindset, they might have considered the neurological
consequences of being exposed to too much blue light. Certain photoreceptive
retinal cells detect blue light, which is plentiful in sunlight. As it turns
out, the brain interprets LEDs much like it interprets sunlight, in effect
telling us it’s time to wake up, making it difficult to get to sleep.</p>
<p class="MsoNormal" style="line-height: 150%;">Problems
like this only serve to emphasize what materials scientist <a href="https://issues.org/impacts-innovations-blue-led-light-ainissa-ramirez/">Ainissa
Ramirez, PhD </a> discusses in a recent essay: “The culture of
research … does not incentivize looking beyond one’s own discipline … Academic
silos barricade us from thinking broadly and holistically. In materials
science, students are often taught that the key criteria for materials
selection are limited to cost, availability, and the ease of manufacturing. The
ethical dimension of a materials innovation is generally set aside as an
elective class in accredited engineering schools. But thinking about the
impacts of one’s work should be neither optional nor an afterthought.”</p>
<p class="MsoNormal" style="line-height: 150%;">This is
the same problem we face in digital health. Too many data scientists and
venture capitalists have invested time and resources into developing impressive
algorithms capable of screening for disease and improving its treatment. But some
have failed to take a closer look at the data sets upon which these digital
tools are built, data sets that misrepresent the populations they are trying to
serve. The result has been an ethical dilemma that needs our immediate
attention.</p>
<p class="MsoNormal" style="line-height: 150%;">Consider
the evidence: <a href="https://science.sciencemag.org/content/366/6464/447">A large commercially
available risk prediction data set</a> used to guide
healthcare decisions has been analyzed to find out how equitable it is. The
data set was designed to determine which patients require more than the usual
attention because of their complex needs. Zaid Obermeyer from the School of
Public Health at the University of California, Berkley, and his colleagues
looked at over 43,000 White and about 6,000 Black primary care patients in the
data set and discovered that when Blacks were assigned to the same level of
risk as Whites by the algorithm based on the data set, they were actually
sicker than their White counterparts. How did this racial bias creep into the
algorithm? Obermeyer et al explain: “Bias occurs because the algorithm uses
health costs as a proxy for health needs. Less money is spent on Black patients
who have the same level of need, and the algorithm thus falsely concludes that
Black patients are healthier than equally sick White patients.”</p>
<p class="MsoNormal" style="line-height: 150%;">Similarly,
evidence from an <a href="https://www.pnas.org/content/117/23/12592">Argentinian
study</a> that analyzed data from deep neural networks
used on publicly available X-ray image datasets intended to help diagnose
thoracic diseases revealed inequities. When the investigators compared
gender-imbalanced datasets to datasets in which males and females were equally
represented, they found that, “with a 25%/75% imbalance ratio, the average
performance across all diseases in the minority class is significantly lower
than a model trained with a perfectly balanced dataset.” Their analysis
concluded that datasets that underrepresent one gender results in biased
classifiers, which in turn may lead to misclassification of pathology in the
minority group.</p>
<p class="MsoNormal" style="line-height: 150%;">These
disparities not only re-emphasize the need for technologists, clinicians, and
ethicists to work together, they beg the question: How can we fix the problem
now? Working from the assumption that any problem this complex needs to be
precisely measured before it can be rectified, Mayo Clinic, Duke School of
Medicine, and Optum/Change Healthcare are currently analyzing a massive data
set with more than 35 billion healthcare events and<span style="background: white;"> about </span><span style="background: white;">16 billion encounters that are linked to data
sets that include social determinants of health. That will enable us to stratify
the data by race/ethnicity, income, geolocation, education, and the like</span>. Creating a platform that systematically evaluates
commercially available algorithms for fairness and accuracy is another tactic worth
considering. Such a platform would create “food label” style data cards that
include the essential features of each digital tool, including its input data
sources and types, validation protocols, population composition, and
performance metrics. There are also several analytical tools specifically designed
to detect algorithmic bias, including Google’s TCAV, Audit-AI, and IBM’s AI-
Fairness 360.</p><p class="MsoNormal" style="line-height: 150%;">The
fences that divide healthcare can be torn down. It just takes determination and
enough craziness to believe it can be done — and lots of hard work.</p>John Halamkahttp://www.blogger.com/profile/04550236129132159307noreply@blogger.com0tag:blogger.com,1999:blog-4384692836709903146.post-37767974751204893722021-07-12T09:22:00.001-07:002021-07-12T15:10:33.903-07:00Identifying the Best De-Identification Protocols<h4 style="text-align: left;">Keeping
patient data private remains one of the biggest challenges in healthcare. A
recently developed algorithm from nference is helping address the problem.</h4><div><p class="MsoNormal" style="line-height: 150%;"><span style="color: black;"><span style="font-family: inherit;"><i>John
Halamka, M.D., president, Mayo Clinic Platform, and Paul Cerrato, senior
research analyst and communications specialist, Mayo Clinic Platform, wrote
this article.</i></span><o:p></o:p></span></p><p class="MsoNormal" style="line-height: 150%;"><span style="font-family: inherit;">In the United States, healthcare
organizations that manage or store personal health information (PHI) are
required by law to keep that data secure and private. Ignoring that law, as spelled
out in the HIPAA regulations, has cost several providers and insurers millions
of dollars in fines, and serious damage to their reputations. HIPAA offers 2
acceptable ways to keep PHI safe: Certification by a recognized expert and the
Safe Harbor approach, which requires organizations to hide 18 identifiers in
patient records so that unauthorized users cannot identify patients. At Mayo
Clinic, however, we believe we must do more.</span></p><p class="MsoNormal" style="line-height: 150%;"><span style="font-family: inherit;">In partnership with the data
analytics firm nference, we have developed a <a href="https://www.sciencedirect.com/science/article/pii/S2666389921000817">de-identification
approach</a> that takes patient privacy to the next level, using a protocol on
EHR clinical notes that includes attention-based deep learning models, rule-based
methods, and heuristics. Murugadoss et al explain that “rule-based systems use
pattern matching rules, regular expressions, and dictionary and
public database look-ups to identify PII [personally identifiable information] elements.” The problem with relying solely on such rules is they miss things,
especially in an EHR’s narrative notes, which often use non-standard
expressions, including unusual spellings, typographic errors and the like. Such
rules also consume a great deal of time to manually create. Similarly, traditional machine learning based
systems, which may rely on support vector machine or conditional random fields,
have their shortcomings and tend to remain reliable across data sets.</span></p><p class="MsoNormal" style="line-height: 150%;"><span style="font-family: inherit;">The ensemble approach used at Mayo<span style="background: white; line-height: 150%;"> includes
</span><span style="background: white; line-height: 150%;">a </span>next generation algorithm that incorporates natural language
processing and machine learning. Upon
detection of PHI, the system transforms detected identifiers into plausible,
though fictional, surrogates to further obfuscate any leaked identifier. We evaluated
the system with a publicly available dataset of 515 notes from the I2B2 2014
de-identification challenge and a dataset of 10,000 notes from Mayo Clinic. We
compared our approach with other existing tools considered best-in-class. The
results indicated a recall of 0.992 and 0.994 and a precision of 0.979 and
0.967 on the I2B2 and the Mayo Clinic data, respectively.</span></p><p class="MsoNormal" style="line-height: 150%;"><span style="font-family: inherit;">While this protocol has many
advantages over older systems, it’s only one component of a more comprehensive
system used at Mayo to keep patient data private and secure. Experience has shown us that de-identified
PHI, once released to the public, can sometimes be re-identified if a bad actor
decides to compare these records to other publicly available data sets. There
may be obscure variants within the data that humans can interpret as PHI but algorithms will not. For
example, a computer algorithm expects phone numbers to be in the form area
code, prefix, suffice i.e. (800) 555-1212. What if a phone number is manually
recorded into a note as 80055 51212? A human might dial that number to
re-identify the record. Further we expect dates to be in the form mm/dd/yyyy.
What if a date of birth is manually typed into a note as 2104Febr (meaning
02/04/2021)? An algorithm might miss that.</span></p><p class="MsoNormal" style="line-height: 150%;"><span style="font-family: inherit;">With these risks in mind, Mayo
Clinic is using a multi-layered defense referred to as data behind glass. The
concept of data behind glass is that the de-identified data is stored in an
encrypted container, always under control of Mayo Clinic Cloud. Authorized cloud
sub-tenants can be granted access such that their tools can access the
de-identified data for algorithm development, but no data can be taken out of
the container. This prevents prevents merging the data with other external data
sources.</span></p><p class="MsoNormal" style="line-height: 150%;"><span style="font-family: inherit;">At Mayo Clinic, the patient always
comes first, so we have committed to continuously adopt novel technologies that
keep information private.</span></p></div>John Halamkahttp://www.blogger.com/profile/04550236129132159307noreply@blogger.com0tag:blogger.com,1999:blog-4384692836709903146.post-63310239050782805322021-07-06T12:31:00.000-07:002021-07-06T12:31:47.165-07:00Learning from AI’s Failures<h4 style="text-align: left;"><span style="color: #333333;">A detailed picture of AI’s mistakes is the canvas upon which we create
better digital solutions.</span></h4><div><p class="MsoNormal" style="line-height: 150%;"><i><span style="color: #333333; mso-bidi-font-family: Calibri; mso-bidi-theme-font: minor-latin; mso-fareast-font-family: "Times New Roman";">John Halamka, M.D., president, Mayo Clinic Platform, and Paul
Cerrato, senior research analyst and communications specialist, Mayo Clinic
Platform, wrote this article.</span></i></p><p class="MsoNormal" style="line-height: 150%;">We all tend to ignore clichés because
we’ve heard them so often, but some clichés are worth repeating. “We learn more
from failure than success” comes to mind. While it may be overused, it
nonetheless conveys an important truth for anyone involved in digital health. Two
types of failures are worth closer scrutiny: algorithms that claim to improve
diagnosis or treatment but fall short for lack of evidence or fairness; and
failure to convince clinicians in community practice that evidence-based
algorithms are worth using.</p><p class="MsoNormal" style="line-height: 150%;">As we mentioned in <a href="http://geekdoctor.blogspot.com/2021/05/ai-enhanced-cardiology-takes-another.html">an earlier
column</a>, <span style="background: white; color: #333333;">a growing number of thought leaders in medicine have
criticized the rush to generate AI-based algorithms because many lack the solid
scientific foundation required to justify their use in direct patient care.
Among the criticisms being leveled at AI developers are concerns about
algorithms derived from a dataset that is not validated with a second, external
dataset, overreliance on retrospective analysis, lack of generalizability, and
various types of bias. A critical look at the hundreds of healthcare-related
digital tools that are now coming to market indicates the need for more
scrutiny, and the creation of a set of standards to help clinicians and other
decision makers separate useful tools from junk science. </span></p><p class="MsoNormal" style="line-height: 150%;"><span style="background: white; color: #333333;">The digital health marketplace is crowded with
attention-getting tools. Among 59 FDA-approved medical devices that
incorporated some form of machine learning, </span><a href="https://pubmed.ncbi.nlm.nih.gov/33853863/"><span style="background: white; mso-bidi-font-family: Calibri; mso-bidi-theme-font: minor-latin; mso-fareast-font-family: "Times New Roman";">49 unique devices</span></a><span style="background: white; color: #333333;"> were designed to improve clinical decision
support, most of which are intended to assist with diagnosis or triage. Some
were designed to automatically detect diabetic retinopathy, analyze specific
heart sounds, measure ejection fraction and left ventricular volume, and
quantify lung nodules and liver lesions, to name just a few. Unfortunately, the
evidential support for many recently approved medical devices varies widely.</span></p><p class="MsoNormal" style="line-height: 150%;"><span style="background: white; color: #333333;">Among the AI-based algorithms that has attracted attention is
one designed to help clinicians predict the onset of sepsis. The Epic Sepsis Model (ESM) has been used on
tens of thousands of inpatients to gauge their risk of developing this
life-threatening complication. Part of the Epic EHR system, it is a penalized
logistic regression model that the vendor has tested on over 400,000 patients
in 3 health systems. Unfortunately, because ESM is a proprietary algorithm,
there’s a paucity of information available on the software’s inner workings or
its long-term performance. Investigators from the University of Michigan just
conducted a detailed analysis of the tool among over 27,600 patients and found
it wanting. </span><a href="https://pubmed.ncbi.nlm.nih.gov/34152373/"><span style="background: white; mso-bidi-font-family: Calibri; mso-bidi-theme-font: minor-latin; mso-fareast-font-family: "Times New Roman";">Andrew Wong and his
associates</span></a><span style="background: white; color: #333333;"> found an area under the receiver operating characteristic
curve (AURAC) of only 0.63. Their report states: “</span>The ESM
identified 183 of 2552 patients with sepsis (7%) who did not receive timely
administration of antibiotics, highlighting the low sensitivity of the ESM in
comparison with contemporary clinical practice. The ESM also did not identify
1709 patients with sepsis (67%) despite generating alerts for an ESM score of 6
or higher for 6971 of all 38,455 hospitalized patients (18%), thus creating a
large burden of alert fatigue.” They go on to discuss the far-reaching
implications of their investigation: “The increase and growth in deployment of
proprietary models has led to an underbelly of confidential, non–peer-reviewed
model performance documents that may not accurately reflect real-world model
performance. Owing to the ease of integration within the EHR and loose federal
regulations, hundreds of US hospitals have begun using these algorithms.”</p><p class="MsoNormal" style="line-height: 150%; mso-layout-grid-align: none; text-autospace: none;"><span style="background: white; color: #333333;">Reports like this only serve to amplify the
reservations many clinicians have about trusting AI-based clinical decision
support tools. Unfortunately, they tend to make clinicians not just skeptical but
cynical about all AI-based tools, which is a missed opportunity to improve
patient care. As we pointed on in a </span><a href="https://catalyst.nejm.org/doi/abs/10.1056/CAT.20.0082"><span style="background: white; mso-bidi-font-family: Calibri; mso-bidi-theme-font: minor-latin; mso-fareast-font-family: "Times New Roman";">recent <i>NEJM Catalyst </i>review</span></a><span style="background: white; color: #333333;">, there are several algorithms that are
supported by prospective studies, including a growing number of randomized
controlled trials.</span></p><p class="MsoNormal" style="line-height: 150%;"><span style="background: white; color: #333333;">So how do we get scientifically well-documented
digital health tools into clinicians’ hands and convince them to use them? One
approach is to develop an evaluation system that impartially reviews all the
specs for each product, and generates model cards to provide end users a quick
snapshot of their strengths and weaknesses. But that’s only the first step. By
way of analogy, consider the success of online stores hosted by Walmart or
Amazon. They’ve invested heavily in state of the art supply chains that ensure
their products are available from warehouses as customers demand them. But
without a <i>delivery service</i> that gets
products into customers’ homes quickly and with a minimum of disruption, even
the best products will sit on warehouse shelves. The delivery service has to
seamlessly integrate into customers’ lives. The product has to show up on time,
it has to be the right size garment, in a sturdy box, and so on. Similarly, the
best diagnostic and predictive algorithms have to be delivered with careful
forethought and insight, which requires </span><a href="https://www.nature.com/articles/s41746-020-00318-y"><span style="background: white; mso-bidi-font-family: Calibri; mso-bidi-theme-font: minor-latin; mso-fareast-font-family: "Times New Roman";">design thinking, process
improvement, workflow integration, and implementation science.</span></a></p><p class="MsoNormal" style="line-height: 150%;"><a href="https://pubmed.ncbi.nlm.nih.gov/32885053/"><span style="background: white; mso-bidi-font-family: Calibri; mso-bidi-theme-font: minor-latin; mso-fareast-font-family: "Times New Roman";">Ron Li and his colleagues at Stanford University</span></a><span style="background: white; color: #333333;"> describe this
delivery service in detail, emphasizing the need to engage stakeholders from
all related disciplines before even starting algorithm development to look for
potential barriers to implementation. They also suggest the need for “empathy
mapping” to look for potential power inequities among clinician groups who may
be required to use these digital tools. It
is easy to forget that implementing any technological innovation must also take
into account the social and cultural issues unique to the healthcare ecosystem,
and to the individual facility where it is being implemented.</span></p><p class="MsoNormal" style="line-height: 150%;"><span style="background-color: white; color: #333333;">If we are to learn from AI’s failures, we need
to evaluate its products and services more carefully and develop them within an
interdisciplinary environment that respects all stakeholders.</span></p><span style="color: #333333;"></span></div>John Halamkahttp://www.blogger.com/profile/04550236129132159307noreply@blogger.com0tag:blogger.com,1999:blog-4384692836709903146.post-70637643632474290822021-06-28T11:53:00.002-07:002021-07-06T10:54:55.428-07:00A Paradigm Shift in Digital Health<h4 style="line-height: 150%; text-align: left;">Innovation is best scaled when pipelines are replaced with platforms.</h4><div><p class="MsoNormal" style="line-height: 150%;"><i><span style="color: #333333; mso-ascii-font-family: Calibri; mso-ascii-theme-font: minor-latin; mso-bidi-font-family: Calibri; mso-bidi-theme-font: minor-latin; mso-fareast-font-family: "Times New Roman";">John
Halamka, M.D., president, Mayo Clinic Platform, and Paul Cerrato, senior
research analyst and communications specialist, Mayo Clinic Platform, wrote
this article.</span></i></p>
<p class="MsoNormal" style="line-height: 150%;"><span style="mso-ascii-font-family: Calibri; mso-ascii-theme-font: minor-latin; mso-bidi-font-family: Calibri; mso-bidi-theme-font: minor-latin;">The <i>Digital Health Frontier </i>includes
cutting edge predictive analytics, machine learning enhanced algorithms and big
data analytics. But for these innovations to have their full impact on patient
care requires the right strategic and operational foundation. In the past, many
technology-focused organizations have relied on a pipeline approach as the
foundation to construct innovations and promote growth. But history suggests
this approach is less sustainable than a platform approach. A recent article in
<a href="https://hbr.org/2016/04/pipelines-platforms-and-the-new-rules-of-strategy"><i>Harvard
Business Review</i></a><i> </i>(HBR) sums up the difference: “</span><span style="background: white; color: #282828; letter-spacing: -0.15pt; mso-ascii-font-family: Calibri; mso-ascii-theme-font: minor-latin; mso-bidi-font-family: Calibri; mso-bidi-theme-font: minor-latin; mso-fareast-font-family: "Times New Roman";">Platform businesses bring together producers and consumers in high-value
exchanges. Their chief assets are information and interactions, which together
are also the source of the value they create and their competitive advantage….
Pipeline businesses create value by controlling a linear series of activities — the
classic value-chain model. Inputs at one end of the chain (say, materials from
suppliers) undergo a series of steps that transform them into an output that’s
worth more: the finished product.” While this explanation gets the point
across, it’s rather abstract. To really appreciate the advantages of one
approach over the other, we need an example or two.</span></p><p class="MsoNormal" style="line-height: 150%;"><span style="background-color: white; color: #282828; letter-spacing: -0.15pt;">Apple’s handset business follows the
pipeline model, making sure there are adequate supplies available to build the
device and then overseeing the various other steps to create a finished iPhone,
as well as its distribution, sales, and servicing. But when Apple linked the
phone to its App store, the situation changed dramatically, turning the
operation into a sustainable platform that connected app developers with iPhone
owners. In their </span><i style="color: #282828; letter-spacing: -0.15pt;">HBR </i><span style="background-color: white; color: #282828; letter-spacing: -0.15pt;">article, Geoffrey Parke and Sangeet Paul Choudary
explain: “The resource-based view of competition holds that firms gain
advantage by controlling scarce and valuable — ideally, inimitable — assets. In a
pipeline world, those include tangible assets such as mines and real estate and
intangible assets like intellectual property. With platforms, the assets that
are hard to copy are the community and the resources its members own and
contribute, be they rooms or cars or ideas and information.” Apple’s success
and the loss of market share by pipeline-oriented companies like Nokia can be
explained by such differences.</span></p><p class="MsoNormal" style="line-height: 150%;"><span style="background: white; color: #282828; letter-spacing: -0.15pt;">Like Apple, </span><a href="https://digital.hbs.edu/platform-digit/submission/farm-to-data-table-john-deere-and-data-in-precision-agriculture/"><span style="background: white; letter-spacing: -0.15pt; mso-fareast-font-family: "Times New Roman";">John Deere has successfully employed a platform approach</span></a><span style="background: white; color: #282828; letter-spacing: -0.15pt;">. They own not just the physical assets — e.g. tractors and combines — but a
vast collection of intellectual property — including APIs and apps to help
farmers manage what is now being called precision agriculture. The company links third party providers and
producers to their farming customers and reaps the benefits. With all these
technological tools in place, farmers now have the ability to more efficiently
monitor their equipment with data on fuel consumption, location, machine hours,
and engine RPMs; and they can improve crop management with weather prediction
data, community pricing and the like. Some of John Deere’s more advanced
combines incorporate a grain quality camera, grain loss sensor, a Gen4 display
monitor, and remote access to an operations center from inside the cab. </span><span style="background: white;">For every new connected tractor sold, more data flows into
the John Deere Platform, enhancing the value of the platform to partners
creating new apps and analytics.</span></p><p class="MsoNormal" style="line-height: 150%;"><span style="background: white;">Mayo
Clinic Platform (MCP) is taking a similar approach. Instead of creating dozens
of pipeline businesses or building an organization chart to support pipeline
businesses, we are leveraging external collaborators, network effects, and data
flowing back to the Platform, which increases its value for producers of
products and consumers of services.</span><b><i><span style="color: #545e6c; mso-ascii-font-family: Calibri; mso-ascii-theme-font: minor-latin; mso-bidi-font-family: Calibri; mso-bidi-theme-font: minor-latin;"> </span></i></b><span style="color: #545e6c;">Mayo Clinic and Commure, a
General Catalyst portfolio health care technology company, have launched Lucem
Health to connect data from remote medical devices with AI-enabled algorithms.<b><i>
</i></b></span><span style="background: white;">External collaborators who have partners
with MCP include nference, Medically Home, Kaiser Permanente, and K Health. The
strategic approach allows
the Mayo Clinic Platform to offer products and services that fall into 4 broad
categories of functionality: Gather, Discover, Validate, and Deliver. For
example, in the Deliver “bucket” is the combined ECG/algorithm system that was
recently validated and published in </span><a href="https://www.nature.com/articles/s41591-021-01335-4"><i><span style="background: white; mso-fareast-font-family: "Times New Roman";">Nature
Medicine.</span></i></a><i><span style="background: white;"> </span></i><span style="background: white;">The digital tool was able to detect low ejection fraction,
thereby improving the diagnosis of </span><span style="background: white; color: #333333;">left ventricular systolic dysfunction. While the ECG/algorithm is improving direct
patient care at Mayo Clinic, it can also be offered to external partners and
embedded in their ECG waive form viewer, which in turn will improve the
relationship between a community hospital and its patient population. </span> <span style="background: white;">Similarly, the clinical data analytics
tools developed by MCP are being made available to outside partners like </span><a href="https://khealth.com/"><span style="background: white; mso-fareast-font-family: "Times New Roman";">K Health</span></a><span style="background: white;">, which provides symptom checking, access to virtual visits
with a clinician. The data analytics function is helping K Health improve its
services to their clientele.</span></p><p class="MsoNormal" style="line-height: 150%;"><span style="background: white;">As the <i>HBR</i> article emphasized, the</span><span style="background: white; color: #282828; letter-spacing: -0.15pt;"> chief assets of a platform are information and interactions, which
together are also the source of the value they create and their competitive
advantage. Such value and advantages are
what will sustain healthcare innovators through the next several decades.</span></p></div>John Halamkahttp://www.blogger.com/profile/04550236129132159307noreply@blogger.com0tag:blogger.com,1999:blog-4384692836709903146.post-6190379926121346802021-06-15T08:42:00.000-07:002021-06-15T08:42:02.504-07:00When AI Meets SDOH<h4 style="text-align: left;">Artificial intelligence can help identify and address the
social determinants of health.</h4><div><p class="MsoNormal" style="line-height: 150%;"><i><span style="color: #333333; font-family: "Georgia",serif; font-size: 10pt; line-height: 150%; mso-bidi-font-family: "Times New Roman"; mso-fareast-font-family: "Times New Roman";">John Halamka, M.D.,
president, Mayo Clinic Platform, and Paul Cerrato, senior research analyst and
communications specialist, Mayo Clinic Platform, wrote this article.</span></i><span style="font-family: "Times New Roman",serif; mso-fareast-font-family: "Times New Roman";"><o:p></o:p></span></p>
<p class="MsoNormal" style="line-height: 150%;">Machine learning is getting better at predicting things.
There are now algorithms that improve the detection of diabetic retinopathy,
predict the onset of sepsis, and help determine a critically ill patient’s risk
of dying. But a piece of wisdom from
Warren Buffet comes to mind: “Predicting rain doesn’t matter. Building arks
does.” Even the most impressive algorithm is relatively useless if it doesn’t
allow us to build better “arks” to address the medical disorder or complications that the digital tool identifies. And building the best healthcare
interventions requires that clinicians not just identify the right signs, symptoms,
and biomarkers, whether they be high cholesterol levels, elevated A1c, or a
lump in a woman’s breast. It requires we
understand what’s happening in patients’ everyday lives <i>outside </i> the clinic, the so-called social determinants
of health (SDOH), and then using that data to inform treatment.</p><p class="MsoNormal" style="line-height: 150%;">A great deal has
been written recently about SDOH. Health professionals are slowly beginning to
realize that we cannot “remove health and illness from the social contexts in
which they are produced,” according to Simukai
Chigudu, Oxford Department of International Development, University of Oxford.<sup>1</sup>
That begs the questions: What social
issues are mostly likely to influence our patients’ clinical course and what do
we do about them? How can AI help alleviate the impact of these issues?</p><p class="MsoNormal" style="line-height: 150%;">The <a href="https://www.cdc.gov/socialdeterminants/tools/index.htm">Centers for
Disease Control and Prevention</a>
(CDC) has numerous data sources to help incorporate SDOH into public health
initiatives and medical practice. But as the agency points out, moving from data
to action is the hard part. CDC has several programs designed to focus
clinicians’ attention on key social issues, including socioeconomic status,
educational level, and work history. <a href="https://www.cdc.gov/niosh/topics/ehr/">One initiative</a>, for instance, zeros in on the role
of EHRs. Its purpose is to support the incorporation and use of structured work
information into health IT systems. How might this SDOH element inform a
physician’s different diagnosis? Consider a patient with hypertension who
doesn’t respond to a low sodium diet or anti-hypertension medication. Awareness
of his 10 year history as a house painter might point the clinician in the direction
of lead poisoning, a possible cause of hypertension. Similarly, a nurse
practitioner may be at a loss to figure out why a patient with type 2 diabetes has recently seen a spike in her A1c
levels. If at EHR system is linked to work
history, when the NP enters the reason for the clinic visit into the EHR field
for chief complaint, this might trigger a pop up box that states that the
patient works the night shift and that shift work can affect diabetes control.
The system would then provide recommendations on diabetes management among
shift workers. The same CDC program is also working on a work information data
model, as well as national standards for vocabulary, system interoperability,
and instructions for health IT system developers. </p><p class="MsoNormal" style="line-height: 150%;">At Mayo Clinic, we are also studying the impact of SDOH
on health and disease. Young Juhn, MD, MPH, Director of the AI Program and
Precision Population Science Lab of Department of Pediatric & Adolescent
Medicine at the Clinic, has studied the
effects of socioeconomic status on health since in 2006 when his research work
was supported by the NIH. With the support from the NIH, he developed and
validated a housing-based socioeconomic measure called the Housing Based Index
of Socioeconomic Status or HOUSES index, which is being used in epidemiologic
research to help understand health disparities and differences in a variety of
health outcomes in both adults and children. The index has enabled researchers
to overcome the absence of socioeconomic measures in commonly used data sources
(e.g., medical records or administrative data), conduct geospatial analysis in
health disparities research, and apply a life course approach.</p><p class="MsoNormal" style="line-height: 150%;">The HOUSES index is an objective way to measure the
individual-level socioeconomic status of
patients because it is based on real property data for individual (not
aggregated) housing units and is derived from public records; it uses 4 data
points: the number of bedrooms in a person’s residence, as well as the number
of bathrooms, square footage of the unit, and estimated building value of the
unit. The index can help target patients who are most at risk of poor health
outcomes and inadequate access to health care , demonstrating the real value of
adding SDOH into the mix by addressing the limitations of the existing SDOH.
For example, Stevens et al have shown that patients with a higher HOUSES score
(quartiles 2-4) had 53% lower risk of
kidney transplant rejection (adjusted hazard ratio 0.47), when compared
to those with the lowest score (quartile 1).<sup>2</sup> Dr Juhn and his
colleagues have found that HOUSES can
predict 44 different health outcomes and behavioral risk factors in both
adults and children.</p><p class="MsoNormal" style="line-height: 150%;">Of course, clinicians still have to be reasonable in
their expectations. Even if an algorithm were outfitted with every conceivable
SDOH, it still may not reduce disparities in healthcare. Patients and providers
may choose to ignore the recommendations of the improved algorithm because they
believe the recommended diagnostic test is too expensive or unjustified, for
example, because it is too difficult for patients to get to the testing
facility, or because a patient’s lack of health literacy prevents them from
seeing the value of said test.</p><p class="MsoNormal" style="line-height: 150%;">Despite these shortcomings, SDOH-enhanced algorithms have
the potential to improve patient care. While physicians and nurses have gained
tremendous insights into health and disease by measuring countless clinical
parameters during office visits, it’s clear now that’s not enough. The clinical picture generated with these
metrics is too often hazy and needs to be supplemented by a long list of social
metrics that can influence a patient’s access to care and their long-term
outcomes.</p><p class="MsoNormal" style="line-height: 150%;"><br /></p>
<p class="MsoNormal" style="line-height: 150%;"><b><span style="color: black; mso-themecolor: text1;">References<o:p></o:p></span></b></p>
<p class="MsoNormal" style="line-height: 150%;"><span style="color: black; mso-themecolor: text1;"><o:p> 1. </o:p></span>Chigudu
S. Book: An ironic guide to colonialism in global health. <i>Lancet. </i>2021.
397:1874-1975.</p>
<p class="MsoNormal" style="line-height: 150%;"><span style="color: black; mso-themecolor: text1;"><o:p> 2. </o:p></span>Stevens
M, Beebe TJ, Wi Chung-II et al. HOUSES index as an innovative socioeconomic
measure predicts graft failure among kidney transplant recipients. <i>Transplantation
</i>2020; 104:2383-2392.</p></div>John Halamkahttp://www.blogger.com/profile/04550236129132159307noreply@blogger.com0tag:blogger.com,1999:blog-4384692836709903146.post-41643742245915087792021-06-11T16:03:00.000-07:002021-06-11T16:03:45.211-07:00The Digital Reconstruction of Healthcare is Upon Us<h4 style="text-align: left;">The transition from brick and
mortar to digital medicine will profoundly impact the way clinicians and
patients interact—and will likely improve clinical outcomes.</h4><p class="MsoNormal" style="line-height: 150%;"><o:p></o:p></p>
<p class="MsoNormal" style="line-height: 150%;"><i>John Halamka, M.D., president, Mayo
Clinic Platform, and Paul Cerrato, senior research analyst and communications
specialist, Mayo Clinic Platform, wrote this article.</i></p><p class="MsoNormal" style="line-height: 150%;"><o:p></o:p></p>
<p class="MsoNormal" style="line-height: 150%;">Paul Cerrato and I are excited to
finally announce the publication of our 5th book together: <a href="https://www.routledge.com/The-Digital-Reconstruction-of-Healthcare-Transitioning-from-Brick-and-Mortar/Cerrato-Halamka/p/book/9780367555979"><i>The
Digital Reconstruction of </i> <i>Healthcare:
Transitioning from Brick and Mortar to Virtual Care</i></a><i>. </i>In March,
we posted the <a href="http://geekdoctor.blogspot.com/2021/03/when-technology-policy-and-urgency-to.html">table
of contents</a> of the new book. Now that it’s reached the “newsstand,” we
wanted to whet readers’ appetite by sharing some additional excerpts.</p><p class="MsoNormal" style="line-height: 150%;"><o:p></o:p></p>
<p class="MsoNormal" style="line-height: 150%;">The logical place to start any
discussion on this topic is to explain why digital reconstruction is necessary, which we
address in Chapter 1:<br /><br /></p><p class="MsoNormal" style="line-height: 150%;"><o:p></o:p></p>
<p class="MsoNormal" style="line-height: 150%;"><b>Episodic Medical Care Often
Falls Short</b></p>
<p class="MsoNormal" style="line-height: 150%;">White coat hypertension, the
tendency for patients to only present with elevated blood pressure during a
doctor visit, illustrates a problem that permeates the entire healthcare
ecosystem. Any sign or symptom that a patient exhibits during an office or
clinic visit may not be a true presentation of their underlying condition.
Unfortunately, this phenomenon not only affects a person’s blood pressure but
other common parameters. White coat hyperglycemia has also been documented. And
since psychosocial stress is likely a contributing cause of such white coat
reactions, white coat hypercholesterolemia, asthma attacks, and numerous other
conditions probably exist as well, all triggered by stress hormones.
Conversely, any normal readings during a physical examination or laboratory
test do not necessarily mean a patient is in good health.</p><p class="MsoNormal" style="line-height: 150%;">The common denominator in all these
scenarios is <i>episodic care</i>. In such situations, clinicians are making a
judgement about a patient’s health status based on cross-sectional data, which
can be misleading. But given the financial restraints and incentives that exist
in healthcare today, it has been the only viable option—until now. With the
emergence of virtual care and remote patient monitoring (RPM), gathering
long-term data for many clinical parameters is no longer out of reach. That steady
stream of online data can be inserted into predictive analytics algorithms to help locate
patients at high risk. Some thought leaders refer to this shift in priorities
as the movement from episodic to life-based care.</p><p class="MsoNormal" style="line-height: 150%;">Such digitally enhanced patient engagement
is the future of healthcare. No responsible practitioner would conclude a
diabetic patient is in good metabolic control based on a single blood glucose
reading, and yet that is often the same reasoning we use when a routine
metabolic panel comes back stating LDL cholesterol, serum calcium, white blood
count, blood pressure, and numerous other parameters are all “within reference
range.” We now have the technology to move beyond this outdated mindset. That
technology enables us to detect longitudinal patterns of change in
patients’ health status. By way of example: Longitudinal data on systolic blood
pressure has been linked to patients’ risk of cardiovascular disease.<br /><br /></p><p class="MsoNormal" style="line-height: 150%;"><o:p></o:p></p>
<p class="MsoNormal" style="line-height: 150%;"><b>The Power of Remote Patient
Monitoring</b></p>
<p class="MsoNormal" style="line-height: 150%;">Many patients and healthcare
professionals have yet to appreciate the power of remote patient monitoring. When
executed correctly, it can be truly transformative, combining medical self-care,
objective physiological data, and expert advice to improve both preventive and
therapeutic care. And as RPM continues to mature, it has the potential to
completely reinvent healthcare, especially among those motivated patients who
see it as a source of self-empowerment. The power of RPM in the hands of
motivated asthmatic patients was well illustrated in an experiment conducted by
University of Wisconsin and Centers for Disease Control and Prevention
researchers. Using an electronic medication sensor that was attached to
inhalers of 30 patients, Van Sickle et al. tracked patients’ use of inhaled
short-acting bronchodilators for 4 months. To evaluate patients’ health status,
investigators asked them to fill out surveys, including the Asthma Control Test
(ACT). One month into the study, they also received weekly emails that summed
up their medication usage for the preceding week and offered suggestions on how
to comply with the National Asthma Education and Prevention Program guidelines.
No changes were observed in ACT scores after the first month, but they
increased by 1.40 points each month after that. Patients also reported
significant decreases in daytime and nighttime symptoms. They also noted
“increased awareness and understanding of asthma patterns, level of control,
bronchodilator use (timing, location) and triggers, and improved preventive
practices.” That last statement is worth closer inspection.</p><p class="MsoNormal" style="line-height: 150%;"><o:p></o:p></p>
<p class="MsoNormal" style="line-height: 150%;">Very often, patients do not
understand the triggers that cause symptoms, unless they are actually attuned
to subtle changes in their physiology. Providing graphic displays of their
symptoms paired with the medication usage can be eye opening for many patients
who never noticed patterns of use before. These newfound revelations were
summed up by several patients participating in the study:</p>
<p class="MsoNormal" style="line-height: 150%; mso-layout-grid-align: none; text-autospace: none;"><span style="line-height: 150%;"><span style="font-family: inherit;"><i>“I learned that I used my inhaler more
than I remember. I was able to see and relate to my doctor that my asthma is
not under control.’’ Participants also reported that the receipt of information
about the time and location where they used their inhaler helped to highlight
locations and exposures to triggers that led to symptoms. ‘‘I’ve been more keen
to note surroundings when I feel shortness of breath,’’ one participant said.
‘‘It opened my eyes to triggers I wasn’t aware of in the past.’’</i></span></span></p>
<p class="MsoNormal" style="line-height: 150%;">The results of this experiment
highlight 2 important lessons for patients and clinicians, summed up in a few
choice words from Kamal Jethwani, MD, MPH, from Partners HealthCare: “The
future of health is proactive, self-managed wellness. We want to put the onus
back on the person. We’re saying: It’s your health, and I’m no longer your babysitter.”</p>John Halamkahttp://www.blogger.com/profile/04550236129132159307noreply@blogger.com0tag:blogger.com,1999:blog-4384692836709903146.post-49983056972080506912021-05-21T13:40:00.002-07:002021-05-21T13:47:20.976-07:00AI-Enhanced Cardiology Takes Another Step Forward<h4 style="text-align: left;">Combining a convolutional
neural network with routine ECGs detected low ejection fraction, a signpost for
Asymptomatic left ventricular systolic dysfunction</h4><div class="separator" style="clear: both; text-align: center;"><a href="https://1.bp.blogspot.com/-msqW_KzR3v0/YKgcPeVHokI/AAAAAAAAKPw/oMGA0iq6wQ4Q7FKZeTGWGaHgzoMdK0KpQCLcBGAsYHQ/s800/Cardiogram.jpg" imageanchor="1" style="margin-left: 1em; margin-right: 1em;"><img border="0" data-original-height="533" data-original-width="800" height="266" src="https://1.bp.blogspot.com/-msqW_KzR3v0/YKgcPeVHokI/AAAAAAAAKPw/oMGA0iq6wQ4Q7FKZeTGWGaHgzoMdK0KpQCLcBGAsYHQ/w400-h266/Cardiogram.jpg" width="400" /></a></div><div><br /></div>
<p class="MsoNormal" style="line-height: 150%;"><i>John Halamka, M.D., president,
Mayo Clinic Platform, and Paul Cerrato, senior research analyst and
communications specialist, Mayo Clinic Platform, wrote this article.</i></p>
<p class="MsoNormal" style="line-height: 150%;">Asymptomatic left ventricular
systolic dysfunction (ALVSD) may not be the most familiar disorder in medicine,
but it nonetheless increases a patient’s risk of heart failure and death. Unfortunately,
ALVSD is not that easily detected. Characterized by low ejection fraction
(EF) — a measure of how much blood the heart pumps out during each
contraction — it’s readily diagnosed with an echocardiogram. But because the
procedure is expensive, it’s not recommended as routine screening for the
general public. A recently developed AI-enhanced algorithm that’s used in
conjunction with an ECG can identify low EF, one of many advances that will
eventually make machine learning an essential part of every clinician’s “tool
kit.”</p><p class="MsoNormal" style="line-height: 150%;">The new algorithm, a joint
effort between several of Mayo Clinic’s clinical departments and Mayo Clinic
Platform, was <a href="https://www.nature.com/articles/s41591-021-01335-4">published online by <i>Nature Medicine</i></a><i>. </i>The EAGLE trial included over 22,000 patients,
divided into intervention and control groups and managed by 358 clinicians from
45 clinics and hospitals. The algorithm/ECG was used to evaluate patients in
both groups but only those clinicians allocated to the intervention arm had
access to the AI results when deciding whether or not to order an
echocardiogram. In the final analysis, 49.6% of patients whose physicians had
access to the AI data underwent echocardiography, compared to only 38.1% (Odds
ratio 1.63, P< 0.001). Xiaoxi Yao,
with the Kern Center for the Science of Health Care Delivery, Mayo Clinic, and
associates reported that “the
intervention increased the diagnosis of low EF in the overall cohort (1.6% in
the control arm versus 2.1% in the intervention arm) and among those who were
identified as having a high likelihood of low EF.” Using the AI tool enabled primary care
physicians to increase the diagnosis of low EF overall by 32% when compared to
the diagnosis rate among patients who received usual care. In absolute terms,
for every 1,000 patients screened, the AI system generated five new diagnoses
of low EF compared to usual care.</p><p class="MsoNormal" style="line-height: 150%;">Earlier research on the neural
network used to create the AI tool had shown that it’s supported by strong
evidence. A growing number of thought leaders in medicine have criticized the
rush to generate AI-based algorithms because many lack a solid scientific
foundation required to justify their use in direct patient care. Among the
criticisms being leveled at AI developers are concerns about algorithms derived
from a dataset that is not validated with a second, external dataset, overreliance
on retrospective analysis, lack of generalizability, and various types of bias,
issues that we discuss in <a href="https://www.routledge.com/The-Digital-Reconstruction-of-Healthcare-Transitioning-from-Brick-and-Mortar/Cerrato-Halamka/p/book/9780367555979"><i>The Digital Reconstruction of Healthcare</i></a><i>. </i>The EAGLE
trial investigators addressed many of these concerns by testing its algorithm
on more than one patient cohort. <a href="https://www.nature.com/articles/s41591-018-0240-2">An earlier study</a> used the
tool on over 44,000 Mayo Clinic patients to train the convolutional neural
network and then tested it again on an independent group of nearly 53,000
patients. And while this study was retrospective in design, <a href="https://pubmed.ncbi.nlm.nih.gov/30821035/">other studies</a> have confirmed the algorithm’s value in clinical
practice by using a prospective design. The most recent study, cited at the
beginning of our blog, was not only prospective in nature, it was also
pragmatic, which reflects the real world in which clinicians practice. Traditional
randomized controlled trials consume a lot of resources, take a long time to
conduct, and usually include a long list of inclusion and exclusion criteria
for patients to meet. The EAGLE trial, on the other hand, was performed among
patients in everyday practice.</p>John Halamkahttp://www.blogger.com/profile/04550236129132159307noreply@blogger.com0tag:blogger.com,1999:blog-4384692836709903146.post-25291535630031319352021-05-14T09:01:00.000-07:002021-05-14T09:01:03.189-07:00A Unique Partnership Delivers Acute and Holistic Home Care<h4 style="text-align: left;">Mayo Clinic, Kaiser Permanente, and Medically Home join
forces to offer patients the best of both worlds, forging a partnership that
has the potential to redefine the hospital industry.</h4><div><p class="MsoNormal" style="line-height: 150%; text-align: left;"><span style="color: black; mso-themecolor: text1;"><span style="font-family: inherit;"><i>John Halamka, M.D., president, Mayo Clinic Platform, and
Paul Cerrato, senior research analyst and communications specialist, Mayo
Clinic Platform, wrote this article.</i></span></span></p><p class="MsoNormal" style="line-height: 150%; text-align: left;"><span style="font-family: inherit;">One of the problems facing most patients who require
hospital admission is removing their familiar surroundings and emotional
supports. While these resources might be considered less critical than
specialized clinical expertise, there's little doubt that these "</span><span style="font-family: inherit;">less
important" factors play a crucial role in the healing process. Even the
most dedicated nursing staff can never replace having a loved one available
24/7 at home. Nor can the most nutritious hospital food ever replace appetizing
home-cooked meals. Equally important are the familiar wake/sleep cycles that
patients are accustomed to at home, which usually must give way to hospital
routines that demand blood pressure checks at three in the morning.</span></p><p class="MsoNormal" style="line-height: 150%; text-align: left;"><span style="font-family: inherit;">A new <a>partnership</a></span><span class="MsoCommentReference" style="font-family: inherit;"><span style="line-height: 150%;"><a class="msocomanchor" href="file:///C:/Users/m222149/AppData/Local/Microsoft/Windows/INetCache/Content.Outlook/VG12LTAI/Holistic%20Acute%20Care%20at%20HomeREVISED24.edited.docx#_msocom_1" id="_anchor_1" language="JavaScript" name="_msoanchor_1">[FJA1]</a> </span></span><span style="font-family: inherit;"> between </span><a href="https://www.mayoclinic.org/" style="font-family: inherit;">Mayo Clinic</a><span style="font-family: inherit;">, </span><a href="https://about.kaiserpermanente.org/" style="font-family: inherit;">Kaiser Permanente</a><span style="font-family: inherit;">, and </span><a href="https://www.medicallyhome.com/" style="font-family: inherit;">Medically Home</a><span style="font-family: inherit;"> launched recently to expand access to care that
combines the comforts of home with the expertise of hospitalists, helping
patients receive the <i>holistic</i> care needed to speed long-term recovery
and the <i>acute</i> care services to address their immediate medical needs.
Stephen Parodi, MD, executive vice president of the Permanente Foundation,
summed up the philosophy behind the new initiative succinctly, "Treating
patients in their home allows physicians to treat the whole patient. We see
their individual needs and can integrate critical information, such as diet,
physical environments and social determinants of health, into their care plans."</span></p><p class="MsoNormal" style="line-height: 150%; text-align: left;"><span style="font-family: inherit;">The challenge, of course, is how to turn this philosophy
into a cost-effective, safe program. In fact, that transition has already
begun. As we discussed in a </span><a href="http://geekdoctor.blogspot.com/2021/02/high-quality-hospital-care-minus.html" style="font-family: inherit;">previous
blog,</a><span style="font-family: inherit;"> Mayo Clinic launched
its advanced care at home program last summer at Mayo Clinic in Florida and
Mayo Clinic Health System in Eau Claire, Wisconsin, to deliver complex,
comprehensive care and restorative services to qualifying patients in their
homes. These services, which are provided in-person and virtually, include:</span></p><p class="MsoNormal" style="line-height: 150%; text-align: left;"></p><ul style="text-align: left;"><li><span style="font-family: inherit;"><span style="color: black; mso-themecolor: text1;">Infusions.</span></span></li><li><span style="font-family: "Times New Roman", serif; font-size: 12pt; text-indent: -0.25in;">Skilled
nursing.</span></li><li><span style="font-family: "Times New Roman", serif; font-size: 12pt; text-indent: -0.25in;">Medication
delivery.</span></li><li><span style="font-family: "Times New Roman", serif; font-size: 12pt; line-height: 150%; text-indent: -0.25in;">Laboratory
and imaging services.</span></li><li><span style="font-family: "Times New Roman", serif; font-size: 12pt; text-indent: -0.25in;">Behavioral
health.</span></li><li><span style="font-family: "Times New Roman", serif; font-size: 12pt; text-indent: -0.25in;">Rehabilitation
services.</span></li></ul><p></p>
<p class="MsoNormal" style="line-height: 150%; text-align: left;"><span style="color: black; mso-themecolor: text1;"><span style="font-family: inherit;">Similarly, Kaiser Permanente launched its hospital-at-home
program in two regions last year, admitting patients from multiple hospitals
across both its Northern California and Oregon locations. In this model, Kaiser
Permanente has a single medical command center in each region, supporting
multiple hospitals to care for patients longitudinally across their acute and
restorative phases.</span></span></p><p class="MsoNormal" style="line-height: 150%; text-align: left;"><span style="font-family: inherit;">The new partnership will scale Medically Home’s
operations, allowing more providers to offer this unique care model. The model includes
a 24/7 medical command center (Figure 1) staffed with clinicians in regular communication with
a care team in the community that contains EMTs and nurses who provide bedside
care. Among the elements that make the new program unique:</span></p><p class="MsoNormal" style="line-height: 150%; text-align: left;"></p><ul style="text-align: left;"><li><span style="font-family: inherit;">Required
protocols for high-acuity care in the home.</span></li><li><span style="font-family: inherit;">Rapid response
logistics systems and providers of care in the home.</span></li><li><span style="font-family: inherit;">Integrated
communication, monitoring and safety system technology in the home.</span></li><li><span style="font-family: inherit;">A
software platform, the Cesia® Continuum, for orchestrating high-acuity care in
patients’ homes. (Figure 2</span><span style="font-family: inherit;">)</span></li></ul><p></p>
<p class="MsoNormal" style="line-height: 150%; text-align: left;"><span style="font-family: inherit;">One of the problems with writing about a significant
health care event while at the same time being a key player in the event is some
outsiders will question our objectivity and immediately assume we exaggerate
its importance to gain a competitive advantage. The plain truth is this
partnership is not just about Mayo Clinic, Kaiser Permanente, and Medically
Home. The partnership's ultimate goal is to bring better care to patients
across the country and the globe. With that in mind, the program will provide all
the necessary outcomes data, tools, systems, training and technology to enable
the model’s widespread adoption.</span></p><p class="MsoNormal" style="line-height: 150%; text-align: left;"><span style="font-family: inherit;"><b>Figure 1</b></span></p><p class="MsoNormal" style="line-height: 150%; text-align: left;"><span style="font-family: inherit;"></span></p><div class="separator" style="clear: both; text-align: center;"><a href="https://1.bp.blogspot.com/-e0jlV1TUQ2U/YJwtGckwzcI/AAAAAAAAKNk/l5JRaUl9qbAcX4Nw2LJbQmWoJDZYML1OACLcBGAsYHQ/s791/Figure%2B1.png" style="margin-left: 1em; margin-right: 1em;"><img border="0" data-original-height="611" data-original-width="791" height="309" src="https://1.bp.blogspot.com/-e0jlV1TUQ2U/YJwtGckwzcI/AAAAAAAAKNk/l5JRaUl9qbAcX4Nw2LJbQmWoJDZYML1OACLcBGAsYHQ/w400-h309/Figure%2B1.png" width="400" /></a></div><p class="MsoNormal" style="line-height: 150%; text-align: left;"><span style="font-family: inherit;"><br /></span></p><b>Figure 2</b><p></p><p class="MsoNormal" style="line-height: 150%; text-align: left;"><span style="font-family: inherit;"></span></p><div class="separator" style="clear: both; text-align: center;"><a href="https://1.bp.blogspot.com/-nMQ3tWrkz00/YJwtREgklSI/AAAAAAAAKNo/-7gazMgu0MwGd6kCFHTtrPWUL4OYueE_QCLcBGAsYHQ/s791/Figure%2B2.png" style="margin-left: 1em; margin-right: 1em;"><img border="0" data-original-height="611" data-original-width="791" height="309" src="https://1.bp.blogspot.com/-nMQ3tWrkz00/YJwtREgklSI/AAAAAAAAKNo/-7gazMgu0MwGd6kCFHTtrPWUL4OYueE_QCLcBGAsYHQ/w400-h309/Figure%2B2.png" width="400" /></a></div><b><br /></b><p></p></div>John Halamkahttp://www.blogger.com/profile/04550236129132159307noreply@blogger.com0tag:blogger.com,1999:blog-4384692836709903146.post-37505517784068382202021-05-05T13:09:00.002-07:002021-05-05T13:16:17.512-07:00Health Data Privacy Gets the Attention It Deserves<h4 style="text-align: left;">The Partners in Privacy Conference gathered
world-class experts to address some of health care’s most vexing problems.</h4><div class="separator" style="clear: both; text-align: center;"><a href="https://1.bp.blogspot.com/-TXtELJCMf1A/YJL89dBnPAI/AAAAAAAAKLc/Ba5IeKkf3CUJ7bGvvLSRXFwwTg9OB1HIwCLcBGAsYHQ/s1500/data-privacy.jpg" imageanchor="1" style="margin-left: 1em; margin-right: 1em;"><img border="0" data-original-height="791" data-original-width="1500" height="211" src="https://1.bp.blogspot.com/-TXtELJCMf1A/YJL89dBnPAI/AAAAAAAAKLc/Ba5IeKkf3CUJ7bGvvLSRXFwwTg9OB1HIwCLcBGAsYHQ/w400-h211/data-privacy.jpg" width="400" /></a></div><br /><div><i>John Halamka, M.D., president, Mayo
Clinic Platform, and Paul Cerrato, senior research analyst and communications
specialist, Mayo Clinic Platform, wrote this article.</i></div><p class="MsoNormal" style="line-height: 150%;"><o:p></o:p></p>
<p class="MsoNormal" style="line-height: 150%;"><span style="font-family: inherit;">The challenges involved in keeping
patient and consumer health data private may seem daunting. Still, a recent
virtual conference hosted by Mayo Clinic brought together over 80 world-class
experts to address the issues. Their insights are worth a closer look.</span></p><p class="MsoNormal" style="line-height: 150%;"><span style="font-family: inherit;">During his opening remarks at <i>Partners
in Privacy Conference: The Ethical and Responsible Use of Data to Drive Cures</i>
(April 22, 2021). Gianrico Farrugia, M.D., CEO of Mayo Clinic, acknowledged the
delicate balance required to respect the public’s desire to keep its data
confidential and the health care community’s desire to use that data to improve
patient care, “What the right thing is is not a simple question to answer </span><span face="Roboto, arial, sans-serif" style="background-color: white; color: #4d5156; font-size: 14px;">—</span><span style="font-family: inherit;"> it
is complex and can vary in different countries and even under different
circumstances. A person may be less or more willing to share information and
have a different view on data privacy at different times in their lives. What degree of data privacy seems right to a
healthy 40 year is likely not going to be the same for that same person with
advanced cancer or neurodegenerative disease. Moreover, the world of research
and collaboration is changing. There are global opportunities for new partnerships
among medical centers, industry and government that increasingly involve data
sharing.”</span></p><p class="MsoNormal" style="line-height: 150%;"><span style="font-family: inherit;">With these concerns in mind, Dr.
Farrugia introduced the keynote speaker, Micky Tripathi, Ph.D., MPP, the
National Coordinator for Health Information Technology for the U.S. Department
of Health and Human Services. Micky briefly reviewed the achievements of the
HITECH Act and the implementation of EHRs around the country. Still, he also
pointed out that the speed with which this rollout occurred has made us all realize
that “technology has outpaced policy.” One of the ways in which this disconnect
is being addressed is through the 21st Century Cures Act. As of April 5, 2021,
the law now requires that providers, health care information networks, and
technology developers give the public friction-free access to and control of
their health data through apps. But that access also means patients can more
easily share that information with third parties that are not required to
follow the rules spelled out in HIPAA. And even when such data remains within
the confines of a health care provider organization bound by HIPAA regulations,
keeping it private and secure remains a challenge, despite the fact that there
are ground rules on de-identifying it. </span></p><p class="MsoNormal" style="line-height: 150%;"><span style="font-family: inherit;">These challenges were among the
many questions addressed by the four breakout groups that followed Micky’s
presentation. We discussed privacy laws and regulations; state-of-the-art
methods for protecting data privacy used to advance health care; consumer and
patient attitudes about privacy; and balancing privacy protection with the
benefits of research and commercialism. The lessons learned from the conference will
be presented in a white paper that is currently being developed by the thought
leaders involved in the project. But in lieu of that, consider a few takeaways:</span></p><p class="MsoNormal" style="line-height: 150%;"></p><ul style="text-align: left;"><li><span style="text-indent: -0.25in;"><span style="font-family: inherit;">Patient consent will evolve. In the future,
there will be more granular control options; we will likely go from a black and
white consent decision to a few more controls based on the use of the data and the
actors who have access to it.<br /><br /></span></span></li><li><span style="font-family: inherit;">There will be more transparency in data use. That
can take the form of a “nutrition label” type description for every algorithm
that clearly spells out the data used to create it, as well as its performance
and characteristics.<br /><br /></span></li><li><span style="font-family: inherit;">Technology is evolving, with machine learning
and natural language processing accelerating very quickly. It will
fundamentally change how we deliver care, but it will also mean more data being
used for more purposes. It is incumbent upon us to keep that data safe and to
respect patient preferences as we develop algorithms. Fortunately, privacy and
security technology is evolving as well, including advances in
de-identification, encryption, allow lists and tokenization. However, the
conference attendees emphasized that these approaches are imperfect, which
means we will need a <i style="text-indent: -0.25in;">multi-layered</i><span style="text-indent: -0.25in;"> approach.</span></span></li></ul><p></p><p class="MsoListParagraph" style="line-height: 150%; mso-list: l0 level1 lfo1; text-indent: -0.25in;"><o:p></o:p></p>
<p class="MsoNormal" style="line-height: 150%;"><span style="font-family: inherit;">Cris Ross, the CIO at Mayo Clinic,
summed up many of the observations gleaned from the conference: “We came into
this conference wondering whether matters of law, regulation, policy,
technology, and market practice are fully established, or if there’s a need for
further exploration and consensus. We concluded there is a need for more
exploration, and a need for a group like this to help create a consensus, with
the goal of advancing cures with the ethical use of data and preservation of
patient privacy?" Cris emphasized that this virtual meeting demonstrated
that this need exists and we would like to issue a call to other leaders in
this field to advance the agenda. <i>Partners
in Privacy Conference: The Ethical and Responsible Use of Data to Drive Cures </i>was
only the first step in a journey that will require the input and expertise of stakeholders
around the nation and the world.</span></p>John Halamkahttp://www.blogger.com/profile/04550236129132159307noreply@blogger.com0tag:blogger.com,1999:blog-4384692836709903146.post-76632940822826804502021-04-26T15:37:00.001-07:002021-04-26T15:38:09.763-07:00It’s OK to Break the Rules Now and Then<h4 style="text-align: left;">Technological
innovation sometimes requires we take risks — and question the tenets of
evidence-based medicine.</h4><p></p><p class="MsoNormal" style="line-height: 150%;"><span style="color: black; mso-bidi-font-family: Calibri; mso-bidi-theme-font: minor-latin; mso-themecolor: text1;"><i>John
Halamka, M.D., president, Mayo Clinic Platform, and Paul Cerrato, senior
research analyst and communications specialist, Mayo Clinic Platform, wrote
this article.</i></span></p><p class="MsoNormal" style="line-height: 150%;">It’s
challenging at times to know when to follow the rules and “color inside the
lines” and when to ignore those lines and forge ahead. That’s true whether
we’re navigating everyday life, creating new technology, or devising the best
patient care initiatives. Which brings to mind a quote from Elbert Hubbard: “The greatest mistake you can make in life is continually fearing that
you'll make one.” </p><p class="MsoNormal" style="line-height: 150%;">Over
the years, we have discussed the strengths and weaknesses of evidence-based
medicine and randomized controlled trials (RCTs) in several books and articles,
our point being that fear of investing in a treatment approach because it not
supported by the RCT “gold standard” can create a kind of inertia that
ultimately hurts patients.<sup>1,2</sup> And even if we put aside the fear factor that
Hubbard mentions, mounting evidence strongly suggests that an in-depth analysis
of large data sets can supplement <span face="Roboto, arial, sans-serif" style="background-color: white; color: #4d5156; font-size: 14px;">— </span>and in some cases be substituted for <span face="Roboto, arial, sans-serif" style="background-color: white; color: #4d5156; font-size: 14px;">— </span>RCTs to
support the clinical decision-making process.</p><p class="MsoNormal" style="line-height: 150%;">That
doesn’t imply that RCTs should be abandoned.
The list of treatments that have been supported or retired due to a
well-designed RCT is long. For decades, surgeons used radical mastectomy to
treat breast cancer until a controlled trial demonstrated that less disfiguring
alternatives were just as effective in managing the disease.<a href="https://academic.oup.com/ejcts/article/53/5/910/4922686?login=true"><sup>3</sup></a> Similarly, clinicians used to freely prescribe hormone
replacement therapy to women in menopause until the Women’s Health Initiative, also
an RCT, demonstrated that it increases the risk of heart disease, stroke, and
breast cancer. But on the other hand, there have been many recent non-RCT investigations
that have taken advantage of the power of massive data sets and have generated
actionable insights. With the VA Boston Health System, Julia Prentice and her
colleagues, using administrative data from more than 80,000 veterans, have
shown that among patients with Type 2 diabetes, sulfonylurea drugs increased
the risk of dying or being hospitalized when compared to patients on thiazolidinediones.<a href="https://pubmed.ncbi.nlm.nih.gov/25498781/"><sup>4</sup></a> Similarly, David Graham created a stir when he analyzed
the patient records of approximately 1.4 million patients who belonged to
Kaiser Permanente in California. They aimed to determine if rofecoxib (Vioxx)
increased the risk of acute myocardial infarction and sudden cardiac death.
Graham et al. reviewed the equivalent of 2,302,029 person-years of follow-up. They detected 8,142 cases of serious
coronary heart disease (CHD), 2,210 of which were fatal. The odds of developing
CHD among patients taking any dose of the medication were 59% greater than it
was among controls. Among patients who took high doses, namely more than 25 mg
daily, the odds of heart disease were 258% greater. <a href="https://pubmed.ncbi.nlm.nih.gov/15705456/"><sup>5</sup></a></p><p class="MsoNormal" style="line-height: 150%;">More recently, nference, a data
analytics firm with a partnership with Mayo Clinic, spearheaded several
data-intensive studies that did not use the traditional clinical trials
protocol. One study used deep neural networks to evaluate 15.8 million clinical
notes in an EHR from over 30,000 patients who underwent COVID-19 diagnostic
testing.<a href="https://elifesciences.org/articles/58227"><sup>6</sup></a> When investigators compared patients with clinically
apparent COVID-19 with negative patients about a week before they had PCR
testing to confirm the diagnoses, they found loss of taste and smell was more
than 37-fold more likely to occur in those whose infection was confirmed versus
those who tested negative. Shweta et al. state, “<span style="color: #222222;">This
study introduces an augmented intelligence platform for the real-time synthesis
of institutional knowledge captured in EHRs.</span>”<span style="color: #222222;"> One caveat that the researchers acknowledge
in the report was that they had yet to conduct prospective validation of the
augmented EHR curation approach.</span></p><p class="MsoNormal" style="line-height: 150%;">A
second nference-based investigation reviewed the records of patients who had
received more than 94,000 doses of the Pfizer COVID-19 vaccine, more than 36,000
doses of the Moderna vaccine, and 1,745 doses of the Johnson & Johnson
vaccine. The study’s goal was to determine the incidence of cerebral venous
sinus thrombosis (CVST), which has been reported in a small number of patients
after receiving one of the vaccines.<a href="https://www.medrxiv.org/content/10.1101/2021.04.20.21255806v1"><sup>7</sup></a><sup> </sup>The preprint study
found no significant association between any of the vaccines as CVST.</p><p class="MsoNormal" style="line-height: 150%;">One
of the strengths of RCTs is their prospective nature, a design that is more
likely to eliminate confounding variables and bias when compared to
retrospective studies. But at the same time, several RCTs have fallen short
because they were underpowered, resulting in false-negative results. Also, RCTs
are expensive and often require many years to generate results that clinicians
can use at the bedside. On the other hand, retrospective analyses can generate
results much more quickly, and under the right circumstances, can provide
actionable insights and inform the clinical decision-making process.</p><p class="MsoNormal" style="line-height: 150%;">Thomas
Frieden, MD, MPH, a former director of the CDC, has pointed out the real-world
advantages of retrospective cohort studies, which have been used to assess the
prognosis and treatment of various types of cancer. That, in turn, has led to
better treatment protocols. Similarly, such cohort studies have successfully
been used to evaluate survival among pediatric cancer patients and made
clinicians aware of the “increased risk of post-treatment cardiac
complications, enabling better clinical care.”<a href="https://www.nejm.org/doi/full/10.1056/nejmra1614394"><sup>8</sup></a> Frieden summed up the controversy this way, “Elevating
RCTs at the expense of other potentially highly valuable sources of data is
counterproductive. A better approach is to clarify the health outcome being
sought and determine whether existing data are available that can be rigorously
and objectively evaluated, independently of or in comparison with data from
RCTs, or whether new studies (RCT or otherwise) are needed.”</p><p class="MsoNormal" style="line-height: 150%;">When comparing research methodologies,
it’s important to remember that’s it’s not a sports competition; there doesn’t
have to be a clear winner and loser. Big data analytics and RCTs each have
their strengths and weaknesses and can be deployed accordingly. When there's
enough time and resources available to conduct a controlled trial, it is often
the best way to evaluate potentially useful treatment approaches. Still, when
clinicians need to quickly make diagnostic and therapeutic decisions,
especially during an international crisis, we don't always have the luxury of
time.</p>
<p class="MsoNormal" style="line-height: 150%;"><b><span style="mso-bidi-font-family: Calibri; mso-bidi-theme-font: minor-latin;"><o:p> </o:p></span></b></p>
<p class="MsoNormal" style="line-height: 150%;"><b><span style="mso-bidi-font-family: Calibri; mso-bidi-theme-font: minor-latin;">References</span></b></p><p class="MsoNormal" style="line-height: 150%; text-indent: 0px;"><span style="text-indent: -0.25in;">1. Cerrato P,
Halamka J. </span><i style="text-indent: -0.25in;">Realizing the Promise of Precision Medicine. </i><span style="text-indent: -0.25in;">2017, Academic
Press/Elsevier, Cambridge, MA, pp. 87-91.</span></p><p class="MsoNormal" style="line-height: 150%; text-indent: 0px;"><span style="text-indent: -0.25in;">2. Cerrato,
P, Halamka J. </span><i style="text-indent: -0.25in;">The Transformative Power of Mobile Medicine. </i><span style="text-indent: -0.25in;">2019,
Academic Press/Elsevier, Cambridge, MA, pp 57-58.</span></p><p class="MsoNormal" style="line-height: 150%; text-indent: 0px;"><span style="text-indent: -0.25in;">3. Treasure
T, Takkenberg JM. Randomized trials and big data analysis: we need </span>the best of both worlds. <i>Eur J
CardioThoracic Surg. </i>2018; 53:910-914.</p><p class="MsoNormal" style="line-height: 150%; text-indent: 0px;"><span style="text-indent: -0.25in;">4. Prentice
JC, Conlin PR, Gellad WF et al. Capitalizing on Prescribing Pattern Variation
to Compare Medications forType2Diabetes. </span><i style="text-indent: -0.25in;">Value in Health. </i><span style="text-indent: -0.25in;">2014;
17:854-862.</span></p><p class="MsoNormal" style="line-height: 150%; text-indent: 0px;"><span style="text-indent: -0.25in;">5. Graham DJ,
Campen D, Hui R, et al. Risk of acute myocardial infarction and sudden cardiac
death in patients treated with cyclo-oxygenase 2 selective and nonselective
non-steroidal anti-inflammatory drugs: nested case-control study. Lancet
2005;365:475–581.</span></p><p class="MsoNormal" style="line-height: 150%; text-indent: 0px;"><span style="text-indent: -0.25in;">6. Shweta F,
Murugadoss K, Awasthi S el al. Augmented Curation of Unstructured Clinical
Notes from a Massive EHR System Reveals Specific Phenotypic Signature of
Impending COVID-19 Diagnosis. <i>eLife. </i>Published online July 7, 2020. </span><a href="https://elifesciences.org/articles/58227" style="text-indent: -0.25in;">https://elifesciences.org/articles/58227</a></p><p class="MsoNormal" style="line-height: 150%; text-indent: 0px;"><span style="text-indent: -0.25in;">7. Pawlowski
C, Rincon-Hekling J, et al. Cerebral venous sinus thrombosis (CVST) is not
significantly linked to COVID-19 vaccines or non-COVID vaccines in a large
multi-state US health system. <i>medRxiv. </i>2021, April 23. </span><a href="https://www.medrxiv.org/content/10.1101/2021.04.20.21255806v1" style="text-indent: -0.25in;">https://www.medrxiv.org/content/10.1101/2021.04.20.21255806v1</a></p><p class="MsoNormal" style="line-height: 150%; text-indent: 0px;"><span style="text-indent: -0.25in;">8. Frieden
TR. Evidence for Health Decision Making —Beyond Randomized, Controlled Trials. </span><i style="text-indent: -0.25in;">N
Engl. J Med.</i><span style="text-indent: -0.25in;">2017;377:465-475.</span></p><p></p>John Halamkahttp://www.blogger.com/profile/04550236129132159307noreply@blogger.com0tag:blogger.com,1999:blog-4384692836709903146.post-12226308082187495482021-04-19T13:04:00.002-07:002021-04-19T13:04:47.860-07:00Can AI Reinvent Radiation Therapy for Cancer Patients?<p><i><span style="background: white; color: #333333;"><span style="font-family: inherit;">John Halamka, M.D., president, Mayo Clinic
Platform, and Paul Cerrato, senior research analyst and communications
specialist, Mayo Clinic Platform, wrote this article.</span></span></i></p><p><span style="font-family: inherit;">Of all the advances in health care artificial
intelligence (AI), medical imaging is probably the most remarkable success story.
Two prominent examples come to mind: Machine learning has helped improve the screening
and diagnosis of retinal disease and is making inroads in skin cancer
detection. Given these developments, it’s not surprising to find researchers
and clinicians developing the digital tools to improve radiotherapy, which
combines imaging technology with high doses of ionizing radiation, delivered through
a device called a linear accelerator.</span></p><p><span style="font-family: inherit;">Radiotherapy is one of the
most common cancer treatments, used to treat more than half of cancers, yet
this labor-intensive expertise is in short supply. <a href="https://www.sciencedirect.com/science/article/pii/S154614400300125X?via%3Dihub"><sup>1</sup></a><sup>
</sup> The digital tools can meet unmet
patient needs for the treatment and increase the accuracy of the delivered
therapy. </span></p><p><span style="font-family: inherit;">To fully appreciate the impact that
AI-enhanced algorithms have on radiotherapy, it helps first to understand how
the equipment and technology used to deliver radiation to a patient’s tumor
functions. Ionizing radiation achieves its purpose by disrupting cellular DNA, which
prevents cancer cells from growing and dividing, which in turn causes solid
tumors to shrink in size. Unfortunately, the same radiation that disrupts tumor
growth can also have a detrimental effect on nearby healthy tissue, resulting
in various of complications.</span></p><p><span style="font-family: inherit;">To minimize this risk, computerized
programs are employed to outline all the anatomical structures closest to the
tumor to be irradiated so that the electron beam will more precisely target the
tumor and spare the healthy tissues — a procedure called contouring. But there is
significant disagreement among providers on how to perform the procedure. Diana
Lin, with the Department of Radiation Oncology, along with several of her
colleagues, point out that such variation is common and “can affect the
resulting plan quality and patient outcomes.” <a href="https://pubmed.ncbi.nlm.nih.gov/32311418/"><sup>2</sup></a><sup> </sup>Their
systematic review also found that variations in target volume delineation was
responsible for greater treatment toxicity and decreased survival. The medical
literature also reveals that major deviations in target delineation occur in up
to 13% of radiation therapy plans.</span></p><p><span style="font-family: inherit;">Although computer programs are available
to help reduce inconsistencies and improve contouring, these digital tools are
far from perfect. <a href="https://newsnetwork.mayoclinic.org/discussion/mayo-clinic-google-launch-ai-initiative-for-radiation-therapy/">Chris
Beltran, Ph.D.</a>, chair of the Division of Medical Physics at Mayo Clinic,
Florida, points out that the relevant organs and tumors “are critical inputs
for the computer models that are currently used to generate radiation dose
plans. If organs are not properly identified, the radiation plan may not
protect these critical structures or adequately treat the tumor.</span>”<span style="font-family: inherit;"> While this
computational modeling reduces the risk to healthy tissue, machine learning is
now being investigated to make contouring more accurate.</span></p><p><span style="font-family: inherit;"><a href="https://newsnetwork.mayoclinic.org/discussion/mayo-clinic-google-launch-ai-initiative-for-radiation-therapy/">Mayo
Clinic and Google Health</a> recently announced a joint initiative focusing on
research into applying AI to radiation therapy planning. Radiation therapy
experts from Mayo Clinic, including radiation oncologists, medical physicists,
dosimetrists and service design, are collaborating with Google Health’s experts
in applying AI to medical imaging. In this first stage of the initiative, Mayo
Clinic and Google Health teams are using deidentified data to develop and
validate an algorithm to automate the contouring of healthy tissue and organs
from tumors and develop adaptive dosage and treatment plans for patients
undergoing radiation therapy for cancers in the head and neck area. The goal of
the IRB-approved project is to develop an algorithm that will improve the quality
of radiation plans and patient outcomes while reducing treatment planning times
and improving the efficiency of radiotherapy practice.</span></p><p><span style="font-family: inherit;">Because the head and neck contain
several sensitive organs that are in close proximity to one another, the Mayo
Clinic/Google project began its investigation in this area of the body. </span>“<span style="font-family: inherit;">Radiation
oncologists today painstakingly draw lines around sensitive organs like eyes,
salivary glands and the spinal cord to make sure radiation beams avoid these areas.
And while this works well, it takes a really long time to get it exactly
right,</span>”<span style="font-family: inherit;"> says Cían Hughes, M.B., Ch.B., informatics lead at Google Health. </span>“<span style="font-family: inherit;">We see huge potential in using AI to augment parts of the contouring
workflow, and hope that this work will ultimately enable a better patient
experience and help patients get the treatment they need sooner.</span>”</p><p><span style="font-family: inherit;">The potentially revolutionary
impact of this new initiative becomes obvious when one considers the fact that
virtually all linear accelerators are equipped with an open-source API, which
means it may be possible for hospitals <i>around the world</i> to use this new
technology to dramatically improve the radiological contouring and making these
treatments available to underserved patient populations.</span></p><p><span style="font-family: inherit;"><br /></span></p><p><b><span style="font-family: inherit;">Reference</span></b></p><p><span style="font-family: inherit;"><span style="background-color: white;">1.<b> </b></span><span style="background-color: white; color: #212121;">Thomadsen
B. The shortage of radiotherapy physicists. J Am Coll Radiol. 2004
Apr;1(4):280-2. doi: 10.1016/j.jacr.2003.12.036. PMID: 17411581.</span></span></p><p><span style="font-family: inherit;"><span style="color: #212121;"><span style="background-color: white;">2. </span></span>Lin D., Lapen, K, Sherer MV et al. A Systematic
Review of Contouring Guidelines in Radiation Oncology: Analysis of Frequency,
Methodology, and Delivery of Consensus Recommendations. <i>Int J Radiology
Oncol. </i>2020; 107:828-835.</span></p><p class="MsoNormal" style="line-height: 150%; mso-layout-grid-align: none; text-autospace: none;"><o:p></o:p></p><p class="MsoNormal" style="line-height: 150%;"><o:p></o:p></p>John Halamkahttp://www.blogger.com/profile/04550236129132159307noreply@blogger.com0tag:blogger.com,1999:blog-4384692836709903146.post-43869172409753642232021-04-09T11:44:00.001-07:002021-04-09T14:01:12.935-07:00A Heart Held Humble Levels and Lights the Way<h4 style="text-align: left;">Observations
on executive training & executive rest</h4><div><i style="background-color: white; color: #333333; font-family: Georgia, serif; font-size: 13px;">John Halamka, M.D., president, Mayo Clinic Platform, and Paul Cerrato, senior research analyst and communications specialist, Mayo Clinic Platform, wrote this article.</i></div><div><p class="MsoNormal" style="line-height: 150%; vertical-align: baseline;"><span style="font-family: inherit;"><span style="color: black; mso-ascii-font-family: Calibri; mso-bidi-font-family: Calibri; mso-fareast-font-family: "Times New Roman"; mso-hansi-font-family: Calibri;">Humility
is not a very popular word among business and health care executives. Often
considered a sign of weakness, this personality trait is not often applauded in
executive training programs or boardrooms. A revealing piece in </span><a href="https://hbr.org/2018/10/if-humility-is-so-important-why-are-leaders-so-arrogant"><i><span style="mso-ascii-font-family: Calibri; mso-bidi-font-family: Calibri; mso-fareast-font-family: "Times New Roman"; mso-hansi-font-family: Calibri;">Harvard Business
Review</span></i></a><span style="color: black; mso-ascii-font-family: Calibri; mso-bidi-font-family: Calibri; mso-fareast-font-family: "Times New Roman"; mso-hansi-font-family: Calibri;"> sums up the problem in its title: “If Humility is So Important, Why
Are Leaders So Arrogant?” The article goes on to discuss the push among HR
consulting firms and psychology experts to develop the H Factor, a combination
of honesty and humility. Despite this celebration of humility, “it
flies in the face of daily headlines in the <i>Wall Street Journal</i> and the
realities of our business and political cultures,” says the HBR article.</span></span></p><p class="MsoNormal" style="line-height: 150%; vertical-align: baseline;"><span style="font-family: inherit;">Several
management experts have tried to explain this paradox. Edgar Shein from MIT
Sloan School of Management posits that the prevailing mindset about managers is
that life is a competition and being a successful leader is all about getting
results at all costs, which in turn requires telling others what to do. There’s
little room for humility and gentleness in that formula for success.</span></p><p class="MsoNormal" style="line-height: 150%; vertical-align: baseline;"><span style="font-family: inherit;">Which
brings us to the blog’s title: a heart held humble levels and lights the way. It’s
a quote from <a href="https://www.google.com/search?q=along+the+road&safe=active&sxsrf=ALeKk03UWSuPgLF3Gkuy3ewOPShP28hZXA%3A1617713583572&source=hp&ei=r1lsYJvtH8Ol_Qah_p_YDQ&iflsig=AINFCbYAAAAAYGxnv6jtRJiFptks6DO7Mwu_XtqpD_BZ&gs_ssp=eJzj4tFP1zcsNM0yS84xKjFg9OJLzMnPS1coyUhVKMpPTAEAjBUJiQ&oq=Along+the+Road&gs_lcp=Cgdnd3Mtd2l6EAMYADIFCC4QkwIyAggAMgIIADICCAAyAggAMgIIADICCC4yAggAMgIILjICCC46BAgjECc6CAgAELEDEIMBOgUIABCxAzoLCC4QsQMQxwEQowI6CAguELEDEIMBOgUILhCxAzoICC4QsQMQkwI6CAguEMcBEK8BUMcLWOojYNU0aABwAHgBgAHcAYgBsAmSAQYxMi4xLjGYAQCgAQGqAQdnd3Mtd2l6&sclient=gws-wiz"><i>Along the Road</i></a>, a
song by Dan Fogelberg. It suggests that informed humility accomplishes two
goals: It levels us, i.e., it provides balance in making decisions, and it lights
up the path as we move forward to accomplish our mission. At the Mayo Clinic
Platform, the pursuit of balance and enlightenment is accompanied by complete
transparency about our goals, dreams, and fears. That certainly requires
humility.</span></p><p class="MsoNormal" style="line-height: 150%; vertical-align: baseline;"><span style="font-family: inherit;">At some
health care organizations, executive coaching is stigmatized. If a leader
falters or is assigned to a role beyond competency (the Peter Principle), a
coach is assigned. We have a different notion. If we're charged with
leading a new team on a new journey with new rules </span>— <span style="font-family: inherit;">the COVID new normal </span>— <span style="font-family: inherit;">we
must embrace the best of what collaborators and partners have to offer. We
think bringing on an external coach as a sounding board during the next six
months of great change will be empowering to us all. The idea is that we'll
meet with the coach twice a month to present our strategy, structure, staffing,
and process ideas to benchmark against the experience of high performing
organizations and teams. It's likely we'll receive feedback and inspiration
that exceeds our own life experiences. At the same time, we'll understand
more about how we can improve efficiency, communications, and decision-making.
The process will be very personal, and we’ll have to grow as people and
leaders. Our life experience has shaped our personalities and our approach to
problems, and although it has served us well in the past, we’ll need to focus
on how to we change to lead the team through the challenges ahead. </span></p><p class="MsoNormal" style="line-height: 150%; vertical-align: baseline;"><span style="font-family: inherit;">Our
life experiences have taught us to take accountability for every situation. As
we build a scalable platform team, it will be more important to orchestrate and
delegate, replacing individual efforts with repeatable processes. Building a
sustainable organization that scales from dozens of projects to hundreds
requires leadership evolution. Coaching can help with such polishing,
especially when working at an accelerated pace. Such coaching is not only
seldomly disclosed, but also rarely documented. However, we’ll keep diaries of
what we learn each month and how it changes our behavior. We'll share
that broadly. Some may suggest that this exposes our vulnerabilities — that’s a
good thing! </span></p><p class="MsoNormal" style="line-height: 150%; vertical-align: baseline;"><span style="font-family: inherit;">Along
the road to informed humility, we also recognize the need for <i>rest. </i>Despite what many executives imagine, the
human body and mind are not perpetual motion machines. The workaholic CEO may
be admired in much of corporate America, but as health professionals, we know
better. The evidence demonstrating the
detrimental effects of overwork on the brain and immune system is overwhelming;
it would be irresponsible for us to ignore it. The American <a href="https://www.stress.org/stress-research">Institute of Stress</a>
calculates that 77% of Americans “regularly experience physical symptoms caused
by stress.” The problem is so prevalent in society, there is even a medical
specialty devoted to it: psychoneuroimmunology, which rests upon one simple
truth: <i>Thoughts have physiological consequences. </i>And ignoring this truth
may not have immediate repercussions for executives, but its insidious effects
eventually take their toll. Solutions abound: Stress management techniques like
mindfulness meditation, walks in the woods, crossword puzzles, long, hot baths,
music — each of us responds to different modalities.</span></p><p class="MsoNormal" style="line-height: 150%; vertical-align: baseline;"><span style="font-family: inherit;">After
all these decades, Fogelberg’s lyrics still offer sage advice to executives who
want to inspire others and serve as role models:</span></p>
<p class="MsoNormal" style="line-height: 150%;"><i><span style="color: #202124; line-height: 150%;"><span style="font-family: inherit;">Along the road<br />
Your steps may tumble<br />
Your thoughts may start to stray<br />
But through it all a heart held humble<br />
Levels and lights your way</span></span></i></p></div>John Halamkahttp://www.blogger.com/profile/04550236129132159307noreply@blogger.com0tag:blogger.com,1999:blog-4384692836709903146.post-2677315439446939192021-03-30T15:52:00.000-07:002021-03-30T15:52:03.197-07:00Health Care Needs Better Marketing, Too<p><i>John Halamka, M.D., president, Mayo Clinic Platform, and
Paul Cerrato, senior research analyst and communications specialist, Mayo
Clinic Platform, wrote this article.</i></p><p><span style="font-family: inherit;">Inspiration comes in all sizes and shapes. Neil deGrasse
Tyson, a world-renowned astrophysicist and director of the Hayden Planetarium
in New York City, continues to inspire us with words like, “The good thing about science is that it's
true whether or not you believe in it.”
Amidst all the confusion and debate in the popular press about health science, this
form of uncommon sense needs more media attention. It’s a truism that may have prompted
Dr. Tyson to pen a recent editorial in <i>The Wall Street Journal</i> entitled “<a href="https://www.wsj.com/articles/neil-degrasse-tyson-on-the-pandemic-year-science-needs-better-marketing-11616106660">Science
Needs Better Marketing</a>.” The same holds
true for health care and the digital tools now available to support it.</span></p><p><span style="font-family: inherit;">We
want the public to appreciate what health science does for us each day, and to
understand that health science is not simply one more opinion that we can
choose to ignore or accept depending on our personal belief system. But how do
we, as scientists, accomplish that heroic feat without alienating consumers and
patients? Or being accused of hyperbole?</span></p><p><span style="font-family: inherit;">Many
consumers see marketing as simply a way to persuade them to buy the latest phone
or fast car. But it can be used for nobler purposes, serving as an essential
tool in creating aware, informed patients. Ansuya Bijur, Director of Marketing
Coordination for Mayo Clinic Platform, points out that, “Patients need to make
informed decisions about their care. Today, the consumer has access to a
plethora of information from their smartphones. A search for ‘are vaccines safe’
on Google gets 1,400,000,000 results. That search includes everything from
academic papers to news stories, social media posts by influencers, to a blog
post by a mom in Kansas City detailing her personal experience. This is where
marketing can help,” explains Bijur. “For example, a year-long marketing
campaign tracking a COVID-19 survivor whose persistent symptoms are treated by
a health care team who find novel therapies/procedures is credible, provides
evidence, is unbiased and balanced.”</span></p><p><span style="font-family: inherit;">Similarly,
there is credible evidence to show that the three vaccines in use are having an
impact of COVID-19 hospitalizations and mortality in the U.S. Those with a
short memory, or are too young to remember, will likely forget that this
accomplishment is only one in a long list that includes the complete
eradication of smallpox, the control of poliomyelitis — after the vaccine was
introduced in 1955, cases dropped from 29,000 to fewer than 900 by 1962 — and numerous
other infections that once killed millions across the globe but now respond to
vaccines.<sup>1</sup> Antibiotics are another victory for health science. Today’s
headlines often feature the dangers of antibiotic overuse and the spread of drug-resistant
microbes. While these concerns need our attention, it would be wrong to let
them distract attention from the millions of lives saved by antibiotics. Anyone
who has lived long enough to remember life in the first decades on the 20<sup>th</sup>
century knows the helplessness one felt when a loved one died from
tuberculosis, a postoperative Staph infection, and countless other killers, or
watched a child with insulin-dependent diabetes disintegrate before their eyes because
there <i>was </i>no insulin available.</span></p><p><span style="font-family: inherit;">Unfortunately,
telling the success stories of medicine seems to have taken a backseat among
many in the mass media to more headline-grabbing negative themes. No doubt,
there are many legitimate stories about medicine’s shortcomings that need to
reach the public. But journalists also need to have the critical thinking
skills to step back on occasion and ask: Am I taking a negative slant because I
know my editor or publisher wants more page views, which in turn translate into
more ad dollars? Am I using needlessly
inflammatory adjectives to spark readers’ anger?</span></p><p><span style="font-family: inherit;">No
one is suggesting that the media roll back the clock to the days when doctors
were viewed as “saints in medical garb” (<a href="https://www.imdb.com/title/tt0638478/">M*A*S*H’s Hawkeye Pierce</a><span class="MsoHyperlink">)</span>.
But today’s culture has moved so far in
the other direction, rewarding a cynicism that goes beyond reason. The comedian Stephen Colbert, in one of his
more serious moments, accurately described the problem, “Cynicism masquerades
as wisdom, but it is the farthest thing from it. Because cynics don’t learn
anything. Because cynicism is a self-imposed blindness, a rejection of the
world because we are afraid it will hurt us or disappoint us. Cynics always say
no. But saying ‘yes’ begins things. Saying ‘yes’ is how things grow.”</span></p><p><span style="font-family: inherit;">Colbert
and Tyson apparently share sentiments. In the <i>WSJ </i>editorial, Tyson says, “… let’s not forget the efforts of lab scientists. Nobody writes stories about
not dying by not contracting COVID-19. So it’s time to praise the researchers
who developed vaccines in record time. If heroes save lives, then they are
superheroes who have saved the lives of millions </span><span style="font-family: source-serif-pro, serif; font-size: 17.816px;">—</span><span style="font-family: inherit;"> because of science.” </span></p><p><span style="font-family: inherit;">Mayo Clinic Platform can add several accomplishments to the list of health care
success stories. In 2020, it launched Advanced Care at Home, a one-of-a-kind
hospital at home program that has managed nearly 1,000 patients to date. And
the Platform is currently finishing the largest-in-history analysis of patient
records to find and weed out systemic racism and other forms of inequality,
using health care data from 50% of the U.S. population. All the evidence points
in one direction: Like many of our colleagues in hospitals and clinics around
the world, we are making a difference in patients’ lives, a difference that more
people should know about!</span></p><p><span style="font-family: inherit;"><br /></span></p><p><b><span style="font-family: inherit;">References</span></b></p><p style="text-indent: 0px;"><span style="font-family: inherit;"><span style="text-indent: -0.25in;"><b>1. </b>Desmond A, Offit
PA. On the shoulders of giants—from Jenner’s cowpox to mRNA Covid vaccines. </span><i style="text-indent: -0.25in;">N
Engl J. Med. </i><span style="text-indent: -0.25in;">2021; 384:1081-1083.</span></span></p>John Halamkahttp://www.blogger.com/profile/04550236129132159307noreply@blogger.com0