Wednesday, January 4, 2017

Our 2017 Priorities

In early 2016, I wrote about the 5 pillars of the BIDMC 2016 work plan - social networking, mobile, analytics, cloud, and keeping data private (SMACK)

As we begin 2017, what should be the focus of our work over the next year?

1. Embrace Social Networking concepts in our workflow, including the EHR

Regardless of the policies, repeals, and delays of the Trump administration, we’ll still need to optimize usability and support the four goals of value-based purchasing - quality measurement, total medical expense management, practice process improvement and technology adoption.   BIDMC has already created a prototype of groupware documentation and we should complete our next generation inpatient documentation solution by mid 2017.    Part of that work incorporates open source secure texting as part of the medical record.   We’re also piloting Google’s G-suite so that our stakeholders can store/share, collaborate, and communicate on any device from anywhere using only a browser.  Our vision is to eventually eliminate the need for a managed desktop.    Google G-suite, Amazon AppStream, Office 365 are all options.

2. Make mobile the preferred form factor for interacting with applications

Mobile is not just a shift in form factor, it’s a shift in user experience that can take into account the location of the user and enable different types of user interaction including voice, touch, and gestures.   Mobile is not just a front end client, it also includes a series of back end services such as APIs, BOTS and machine learning.  What do I mean?   A mobile app that uploads a clinical photograph needs an image exchange API with the EHR.   An app that a patient might use to schedule an appointment online might include a BOT to engage in a dialog about time of day preferences and the severity of illness.  We recently installed a bedside medication dispensing unit that tracks patient reported pain scores and patient self-administration of pain meds.   We’ll be able to use that data to better understand how to keep patients comfortable with less narcotic medication.   Tools we are investigating include Amazon Echo/Alexa, image recognition, text to voice, internet of things integration and artificial intelligence.

3.  Enhance Analytics

If you ask 100 doctors to describe population health and care management automation, you’ll get 150 answers.   Our view is that two kinds of technologies are needed - cohort identification and task list management.     Cohort identification should work with structured and unstructured data, enabling patients to be enrolled in a specific disease/care management program using arbitrarily complex boolean expressions i.e. find all patients who have been described in free text notes as “sad” and examine their medication list for classes of therapeutics with depressive side effects.

Once they are identified as part of cohort, a care plan of arbitrary complexity with tracking of tasks to do/tasks undone should enable a care team to take responsibility and ensure all loops are closed.

We’ve built such a system for several conditions already and we’ll make the tools even more flexible over the next year.

4.  Migrate to the Cloud

BIDMC’s cloud application services will be open for business in 2017.  We’re taking a
careful approach but we’re very much on a trajectory to maximize sensible usage of cloud services.

In past years, cloud services were limited to low risk decision support databases and online educational materials purchased via subscription from content vendors.   That’s analogous to buying the New York Times online from the cloud.   Today we have contracts in place with Amazon Web Services for application hosting/storage/analytics, Google for G-Suite, and Dell/NTTData for community hospital EHR hosting.  Additionally our community ambulatory practices use AthenaHealth’s EHR/practice management from their cloud

We believe we can achieve high reliability, robust geographically distributed disaster recovery, security best practices, lower cost, and enhanced scalability with these services.

Our experience suggests that IT FTEs are unlikely to be increased.   The only way to enhance innovation and customer service is to move the work that can be moved to the cloud, freeing up time of existing IT FTEs for new work.

One of our challenges has been ensuring network latencies over the internet are low.    We created a 10 Gigabit dedicated network connection using AWS's Direct Connect service connection from our network core to Amazon with a latency of 11 ms maximum.  You cannot tell the difference between locally hosted and cloud hosted.

Most mainstream cloud hosting companies will now sign a business associate agreement (BAA), following industry best practices (NIST 800-66, COBIT, ITIL) for protecting patient privacy.   All over our vendors have provided us with  third party audits demonstrating adherence to some reasonable security and privacy framework.

Although many companies are willing to implement privacy controls and sign a BAA, almost no company is willing to indemnify customers for privacy breaches.   We have purchased a third party cyber liability policy.  

5. Keep information private

Security will continue to be a major focus.   In 2017, we’ll roll out  two factor authentication which will serve as a foundation for e-prescribing of controlled substances as well as mitigating the risks of password theft.

We’ll continue our Blockchain experiments as a way to keep data safe and ensure data integrity across multiple organizations as we coordinate care across the continuum of the patient experience.

Those are the 5 pillars for 2017 and the next technologies we’ll explore to support them.

Just as in previous years, I can say that I’m excited by the Healthcare IT adventures ahead.

No comments: