tag:blogger.com,1999:blog-4384692836709903146.post7190481160785076570..comments2024-03-27T09:55:23.143-07:00Comments on Dispatch from the Digital Health Frontier: Freeing the DataJohn Halamkahttp://www.blogger.com/profile/04550236129132159307noreply@blogger.comBlogger8125tag:blogger.com,1999:blog-4384692836709903146.post-36519095530031596362011-03-14T08:30:40.765-07:002011-03-14T08:30:40.765-07:00Just a small comment: all the technology you menti...Just a small comment: all the technology you mention is already out t/here :)<br /><br />Instead, I'd ask you: what are the biggest obstacles to their deployment in a healthcare environment?<br /><br />Thanks, and greetings,<br /><br />Javier TorresDoktor Towersteinhttps://www.blogger.com/profile/18024654269314372646noreply@blogger.comtag:blogger.com,1999:blog-4384692836709903146.post-48116621461097142982011-03-11T10:08:06.373-08:002011-03-11T10:08:06.373-08:00Great post John. I really like the idea of leverag...Great post John. I really like the idea of leveraging unstructured data, I think this is often overlooked but can probably give the user some of the most valuable information. We have bookmarked this post on our community for IM professionals (www.openmethodology.org). Look forward to reading your work in the future.Unknownhttps://www.blogger.com/profile/09956454901557518064noreply@blogger.comtag:blogger.com,1999:blog-4384692836709903146.post-39650656913430067562011-03-03T20:08:36.077-08:002011-03-03T20:08:36.077-08:00Data Liberación!!!!!Data Liberación!!!!!Ahierhttps://www.blogger.com/profile/13398190978662246852noreply@blogger.comtag:blogger.com,1999:blog-4384692836709903146.post-4484921848429689842011-03-02T20:42:03.219-08:002011-03-02T20:42:03.219-08:00It was great to see you in the Apple iPad video.It was great to see you in the Apple iPad video.Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-4384692836709903146.post-65712613286329381122011-03-02T18:32:23.892-08:002011-03-02T18:32:23.892-08:00You are right on the money with talking about how ...You are right on the money with talking about how we design applications for connectivity and thus perhaps we are moving up to the next level again:)<br /><br />Mapping the free text once it gets there will really enhance knowledge and I'm still watching to see how that end of the scheme progresses, like you said to not lose meaning. <br /><br />I liked Watson too and actually found a needier place than healthcare I think:) With speech recognition, connection to the web, it could open doors for our lawmakers to get the timely connected information they need to query too in order to create more effective laws with fewer unintended consequences,(just like we want in healthcare) I would like that as much as anyone out there. <br /><br />I did see where Watson made a cameo appearance but would be nice if it were parked in DC full time:)<br /><br />http://ducknetweb.blogspot.com/2011/03/ibm-watson-did-get-chance-to-go-to.html<br /><br />Natural language processing might be an asset there too as well as in healthcare:)Medical Quackhttps://www.blogger.com/profile/12443589277651479846noreply@blogger.comtag:blogger.com,1999:blog-4384692836709903146.post-19593415812557784722011-03-02T17:51:53.965-08:002011-03-02T17:51:53.965-08:00We are very energized by this blog entry. While we...We are very energized by this blog entry. While we are well-aligned with your vision in general, we have added some specific comments on our experience developing search and reconciliation technology.<br /><br />“1. Both structured and unstructured data will be important”<br /><br />We agree. This short video (www.apixio.com/video-demo) demonstrates how we’ve taken both your personal CCD and unstructured PDF medical records and made them instantly available in a unified, comprehensive view in response to a directed clinical query by the physician (for example “palpitations”). <br /><br />“2. Inference is possible by parsing natural language”<br /><br />This is paramount for increasing search precision and extracting useful information. We find that by applying statistical semantic analysis techniques one can infer context and improve search quality. As a real example, a search for “anxiety” returns the problem “stress” but not the procedure “stress test,” a common cardiology assessment.<br /><br />“3. Data mining needs to go beyond single databases owned by a single organization.“<br /><br />At the HIMSS11 Interoperability Showcase (see our white paper at www.apixio.com/white-papers) we have demonstrated the ability to run a live query across many IHE actors, thereby supporting key clinical information sharing. The Direct Project holds great promise as well.<br /><br />“5. The President's Council of Advisors on Science and Technology (PCAST) Healthcare IT report will offer several important directional themes to will accelerate "freeing the data".<br /><br />We are in favor of a Universal Exchange Language and are ready to contribute to its development. In the meantime, we have found a way to combine and reconcile multi-source data using a suite of natural language processing (NLP) techniques. Perhaps an alternate solution is to develop a "Universal way of Exchanging Language." <br /> <br />“6. Ontologies and data models will be important to support analytics”<br /><br />We too believe that middleware solutions that enable federation of data are the keystone to data access and analytics. Our current solution is to allow any provider or consumer application to query clinical information from our system via REST and SOAP APIs.<br /><br />“7. Mapping free text to controlled vocabularies will be possible and should be done as close to the source of data as possible.”<br /><br />We absolutely agree with you. The clinician does not think about search in terms of SNOMED-CT or ICD-10 codes. We find that our users prefer to search using “doctor-friendly terms” such as “chest pain” to retrieve related information such as CKMB results, “GERD” and “Holter monitor.”<br /><br />While we agree with your vision of associating controlled vocabularies with free text as the data is entered, in today’s environment we find overwhelming amounts of incomplete structured data and diverse textual data. To ensure data integrity in the face of such data heterogeneity, our DisplayMerge technology reconciles and merges data at display time and provides simple ways for users to control quality and change data associations (merges). Changes made by any user persist within the network in a Wiki fashion, allowing errors to be resolved over time.<br /><br />“9. New tools will empower end users”<br /><br />There is a lot of work to be done in this space, but at this point we believe search is the easy button of healthcare information. (John- Next time we meet, we would love to show you our next generation smart search capabilities.)<br /><br />“10. Novel sources of data will be important”<br /><br />We have learned that it is possible to extract medical knowledge from the Web to create associations between medical terms. <br /><br />This is an exciting time in our industry. While today’s technology allows clinical data coming from multiple sources to be instantly accessible and searchable, our biggest challenge lies in how we can expand these capabilities to broad provider and consumer networks, while maintaining privacy and security. <br /><br />We look forward to your keynote presentation and future dialog.<br /><br />Bob Rogers<br />Chief Scientist<br />Apixio Inc.Bob Rogershttp://www.apixio.comnoreply@blogger.comtag:blogger.com,1999:blog-4384692836709903146.post-25547396931171752842011-03-02T08:12:26.807-08:002011-03-02T08:12:26.807-08:00Great post John. I wish I could be there to hear y...Great post John. I wish I could be there to hear your presentation. What you summarize here has been a long time coming - and applicable to virtually every industry. I too have been using the IBM Jeopardy Watson search engine as a mainstream popularized example of what NLP and intelligent search is all about. For several years most would not believe me when I spoke of such things. Watson may do for intelligent search what Google did for "basic" search.<br /><br />I look forward to your posting of your presentation.Carl Frappaolohttp://www.takingaiim.comnoreply@blogger.comtag:blogger.com,1999:blog-4384692836709903146.post-83328440302349611972011-03-02T07:07:20.166-08:002011-03-02T07:07:20.166-08:00When talking about freeing data, atleast at a larg...When talking about freeing data, atleast at a large healthcare facility or IDN level, the idea of Vendor Neutral Archiving continues to be brought up as a relevant and viable solution. What are you thoughts on how VNA and how it fits into the larger picture of "freeing data."Anonymousnoreply@blogger.com