Friday, October 30, 2015

Use Cases for Data Science at Academic Health Science Centers

Like many academic health science centers, my institution is undergoing a planning process to determine our strategy for data science. I have expressed my concern about the (lack of?) differences between data science and biomedical and health informatics, but the former term seems to be carrying the day. I consider it a personal mission to ensure that the long learned history of biomedical and health informatics is not lost in our rush to embrace this seemingly new data science.

One of my major contributions to our process has been to delineate a set of use cases for data science in academic health science centers. These institutions are distinct from organizations that are predominantly devoted to healthcare delivery, and tend to have small or non-existent research and education missions, and general universities, which may not have healthcare delivery activities integrated with their research and educational missions.

I have broken down my use cases into the three general missions that most academic health science centers have. I only present these at a high level, and there is obviously a much greater depth of detail that could be described for each. But these are the big-picture use cases that in my view drive data science in academic health science centers.

Use cases for the clinical mission of academic health science centers include:
  • Clinical decision support – improve clinical practice via predictive analytics and other uses of patient data, including precision medicine as it works its way into clinical practice
  • Quality measurement and improvement – use data to measure and improve quality of care delivered, especially as healthcare shifts to new value-based models of care
  • Business intelligence – apply data to improve business and financial operations of healthcare delivery
  • Patient engagement – patients upload and interact with data related to their care
  • Public health surveillance – use data for early detection and intervention in public health threats (natural and manmade)
Research is also critically important for academic health science centers, and here are some broad use cases, with many important variations on these themes:
  • Prospective studies – improve data capture and analysis for clinical trials and related studies
  • Retrospective studies – enhance ability to use data already collected
  • Basic science research – studies in the "omics," imaging, and other areas that lead to health-related applications
  • "Third science" research – advancing the science of healthcare delivery, the third science of healthcare (after basic and clinical sciences)
  • Data science and informatics research – advance the theory and practice of data science and biomedical and health informatics
Education is also a vital mission for academic health science centers, not only to train users and managers of data but also professionals and researchers who implement and advance the science:
  • Training for data users and managers, clinicians, and others – allowing those who implement programs applying data science to be more savvy in doing so
  • Education for data science and informatics professionals – master's-level education as well as the new clinical informatics fellowships for physicians
  • Advanced education for data science and informatics researchers – doctoral-level education to advance the science of this work
Data science is indeed unique in academic health science centers. These use cases demonstrate how it spans across all of their missions. The success of initiatives such as ours are likely to depend upon the integration all of three.

Tuesday, October 27, 2015

Meeting My Doppelgänger (Googlegänger)

One of my teachable moments in information retrieval (IR) is about uncommon words tending to be the most discriminating and leading to the best results in searching. I am hardly the first person to come up with this idea, as IR research pioneer Gerald Salton demonstrated its value and published about it in the 1970s [1]. I do, however, provide a modern example of it, which is demonstrated by searching (or Googling) on my name. My last name, in particular, is spelled in a somewhat unusual manner, as most people spell it Hirsh, Hersch, or Hirsch. Combined with my presence on the Web, with many links to my major pages (another teachable moment about the Google PageRank algorithm [2]!), I have never had to pay anyone for search engine optimization (SEO), and Googling “Bill Hersh” or “William Hersh” lists most of my key pages right at the top of the search output.

Early in the days of Google, I discovered another William Hersh, who was also in academia. I also noted him in PubMed (MEDLINE) author searches on our name (hersh w). I knew he was a Chemistry professor at Queens College in New York. Apparently over this nearly two decades, he knew of me as well. We both contemplated reaching out to each other, but neither of us ever did.

About a month ago, I received an email from a colleague of his at Queens College that was sent to my Gmail account by mistake. I replied to the email, telling the sender it was sent to me in error and probably meant for his co-worker. He sent my message to Bill, who reached out to me to apologize for the error. This started a conversation, with each of us describing how long we knew about each other. (He was even once invited to serve on an NIH review panel when they thought they were inviting me!). He also told me that a former student of his from years ago, upon finding me, told him that “that my googleganger is my doppelgänger.” (I have to admit I had to go Google the word doppelgänger to be certain of its meaning.)

We both noted we were academic graybeards, and after some discussion found out that our grandfathers emigrated from different cities in Poland (his Czestochowa and mine Lodz). In addition, like many people with last name Hersh, our grandfathers Anglicized their last names from Hershkowitz. They also both experienced anti-semitism in Eastern Europe, part of their motivation for emigrating to the US.

I also told Bill that I was going to be in New York City in late October, and we set a day and time to meet for lunch. We had that meeting yesterday, and it was enjoyable to trade stories of our somewhat common ancestry, our careers, and our families. My family got a kick out of my telling them that Bill too drives a Toyota Prius. Here is a picture of us together:

It was indeed fun to find Bill, meet him, and reflect on how our meeting was made possible by the Web and IR (my field of research). I do hope to keep in touch with him and meet him again.


1. Salton, G, Yang, CS, et al. (1975). A theory of term importance in automatic text analysis. Journal of the American Society for Information Science. 26: 33-44.
2. Brin, S and Page, L (1998). The anatomy of a large-scale hypertextual Web search engine. Computer Networks and ISDN Systems. 30: 107-117.

Friday, October 9, 2015

A Huge Week for Health IT/Informatics

This past week was a busy week in the health IT/informatics world, as the US government released a flurry of rules and documents around health IT. As I tell my students, it is great to be living in this ever-changing part of the history of our field.

Probably making the most news was the release of the rules for Stage 3 of the EHR Incentive (also known as "Meaningful Use") Program by the Centers for Medicare & Medicaid Services (CMS). The Meaningful Use Program has taken its share of lumps in the last year or two, with the challenges providers have had in meeting its Stage 2 criteria and how it has consumed bandwidth that might be put toward other innovation by the healthcare system as well as the vendors. CMS has seem to have gotten the message somewhat, and the new criteria do dial back some on the requirements.

With the new rule, Stage 2 will be modified significantly. Some acute relief will be provided in the form of reduced requirements, from the necessity of reporting only 90 days (as opposed to a full year) of annual reporting to modification of the "view, download, and transmit" (VDT) requirement from five percent of an EP's patient panel to one single patient and reducing the secure messaging requirement from five percent to just being required to have the capability.

Also changed in Stage 2 itself, now called Modified Stage 2, which will be in effect from 2015-2017. The number of objectives is reduced to ten for eligible professionals (EPs) and nine for eligible hospitals (EHs), with each having one or more measures. The objectives are:
  1. Protect Patient Health Information
  2. Clinical Decision Support (CDS)
  3. Computerized Provider Order Entry (CPOE)
  4. Electronic Prescribing
  5. Health Information Exchange
  6. Patient Specific Education
  7. Medication Reconciliation
  8. Patient Electronic Access
  9. Secure Electronic Messaging (EPs only)
  10. Public Health Reporting
Starting in 2018, Stage 3 will become active, with the same objectives as above but with some more rigorous criteria for some of the measures. There is, however, one qualification to Stage 3, which is the opening of a comment period for how it could be changed to align with the new value-based care rules for Medicare. With the addition of calls for Stage 3 to be delayed or outright abandoned, it is not clear what it will ultimately look like.

The full rule is available, as is a brief summary. As always, my preference is for a detailed overview that provides enough detail for the informed reader, somewhere in between the minimally informative short summary and the exhaustive detail of the entire review, which CMS has also provided. (Note to standards developers! I prefer this approach for documentation of standards as well, eschewing both the superficial overviews as well as hundreds-of-pages implementation guides.)

Always a companion to the release of rules for the EHR Incentive Program is the release of the Health Information Technology Certification Criteria by the Office of the National Coordinator for Health IT (ONC). However, as noted by ONC, going forward the EHR Incentive Program will be decoupled from Health IT Certification. The EHR Incentive Program will still required use of certified products, but certification will also be used for other health IT functionality. As with CMS, a short summary and the detailed rule are provided, with the only interim document at this time being a Powerpoint deck that was used in the Webinar ONC presented to describe the new criteria.

The week’s activities did not, however, stop with release of the meaningful use and certification rules. ONC also released a final version of its Federal Health IT Strategic Plan for 2015-2020.
The stated mission of the plan is to "improve the health and well-being of individuals and communities through the use of technology and health information that is accessible when and where it matters most.” This will be achieved through four goals:
  1. Advance Person-centered health and self-management
  2. Transform health care delivery and community health
  3. Foster research, scientific knowledge and innovation
  4. Enhance the nation’s health IT infrastructure
One of the objectives of the fourth goal is to implement the Shared Nationwide Interoperability Roadmap, which was also released by ONC this week in its final Version 1.0 form. The roadmap was accompanied by an updated version of ONC's 2016 Interoperability Standards Advisory, which provides an exhaustive list of the best available standards and links to their implementation specifications. These releases were described in a blog post by ONC Director, Dr. Karen DeSalvo.

As if this week’s activities were not enough, last week was another major milestone, with the switchover to ICD-10-CM by hospitals physician offices, and others who bill in the healthcare system. Eerily similar to Y2K a decade and a half ago, there were very few reports of problems, presumably because the community was well-prepared. Of course, only time will tell, particularly if providers start having claims denied because of faulty coding.

Another recent event pertinent to all of the above occurred the week before, when I presented the inaugural Clinical Informatics Grand Rounds at OHSU. The Grand Rounds series will be part of our normal Thursday Conference Series, and I usually kick off the series each academic year. This year I chose to talk on the topic, HITECH and Meaningful Use: Results from the Grand Experiment and Future Directions. My talk (video and slides available) was built around a proclamation I made in this blog on January 24, 2010, in a posting entitled, Informatics Now Lives in a HITECH World:
"This is a defining moment for the informatics field. Never before has such money and attention been lavished on it. HITECH provides a clear challenge for the field to 'get it right. It will be interesting to look back on this time in the years ahead and see what worked and did not work. Whatever does happen, it is clear that informatics lives in a HITECH world now." Going forward, it will continue to be interesting to pause and reflect.

Thursday, October 1, 2015

Accolades for DMICE

As regular readers of this blog know, I periodically devote postings in this blog to accolades, usually for myself but sometimes for others. I would like to devote this posting to accolades for the many students and faculty in the Oregon Health & Science University (OHSU) Department of Medical Informatics & Clinical Epidemiology (DMICE). More details are provided in the recently published edition of our department newsletter. (Past newsletters are also available.

There are many accolades to point out in the newsletter (starting on page in parentheses):
  • Our new Accreditation Council for Graduate Medical Education (ACGME)-accredited clinical informatics fellowship has launched with its first two fellows (1)
  • Thirty-seven individuals graduated with degrees and certificates in biomedical informatics from OHSU in the 2015 academic year (2)
  • Faculty member Nicole Weiskopf, PhD receiving a Catalyst award from the Oregon Clinical and Translational Research Institute (OCTRI) (3)
  • Numerous faculty and students participating in OHSU Research Week in May, 2015 (4)
  • A surprise party celebrating 25 years at OHSU for myself and fellow faculty Mark Heland, MD (10)
  • New OHSU School of Medicine leadership appointments for Paul Gorman, MD and Heidi Nelson, MD (15)
One final accolade is particularly noteworthy to call out. We were recently informed that OHSU informatics students will be finalists in the 2015 AMIA Student Design Challenge. The theme of this year’s competition is, The Human Side of Big Data – Facilitating Human-Data Interaction. A number of student groups from around the country submitted entries to the competition, and four finalists were recently selected to present at the AMIA Annual Symposium in November. Two of those finalist groups consist of OHSU students:
  • Ashley Choi, Benjamin Cordier, Prerna Das, PhD, and Jason Li, MS will present on, “Take a Breather: Empowering Adherence & Patient Centered Research through Interactive Data Visualization, Social Engagement, & Gamification in Patients with Sleep Apnea.”
  • Michelle Hribar, PhD, L. Nelson Sanchez-Pinto, MD, Kate Fultz Hollis, MS, Gene Ren, and Deborah Woodcock, MBA will present on, “Learning from the Data: Exploring a Hepatocellular Carcinoma Registry Using Visual Analytics to Improve Multidisciplinary Clinical Decision- Making.”
I am delighted that our students were successful enough to get this far, and I hope that one of them emerges as the winner, as a group of OHSU informatics students did in last year’s event.