Demystifying Patient Matching Algorithms
Steven Posnack, M.S., M.H.S. Deputy National Coordinator for Health Information Technology | May 1, 2017
Last week at Health Datapalooza 2017, Adam Culbertson (HIMSS Innovator in Residence at ONC) and I gave a five minute “coming attraction” presentation about a patient matching algorithm challenge ONC will launch in June. For the uninitiated, we use “patient matching” in health IT as shorthand to describe the techniques used to match the data about you held by one health care provider with the data about you held by another (or many others). In practice, patient matching is the process of comparing different demographic elements from different health information technology (health IT) systems to determine if they refer to the same patient.
From an interoperability perspective, the ability to complete patient matching efficiently, accurately, and at scale has long been identified as a key element of the nation’s health IT infrastructure. Patient matching is almost universally needed to enable the interoperability of health data for all kinds of purposes. Patient matching also requires careful consideration with respect to its effect on patient safety and administrative costs.
While numerous recommendations have been issued over the years to tackle different aspects of patient matching, it is important to recognize that the entire health care system can impact its performance – from data capture at patient registration to the technology and algorithms along the way. At the same time, there has been little transparency about how well current patient matching algorithms perform and no industry-accepted minimum baseline(s), benchmark(s) or testing approach(es) exist.
To promote the need for better transparency about the performance of patient matching algorithms, ONC is launching the Patient Matching Algorithm Challenge. We expect the result of this challenge will spur the development of innovative new algorithms, benchmark current performance, and help industry coalesce around common metrics for success.
Up to six cash prizes will be awarded with a total purse of up to $75,000. The major prize category will involve three prizes for the highest “F-Score,” which is the combination of best precision and recall. Additional “best in category” prizes will be awarded for “best precision” (least mismatched patients), “best recall” (least missed matches) and “best first F-Score run.” Participants in the Challenge will get up to 100 tries to score their matching solution.
Prior to the Challenge’s launch in June, ONC will be holding repeat informational webinars on three Wednesday’s in May (May 10th, 17th, and 24th). Visit the Challenge site for more information including dates for informational webinars.
Interested in this challenge? Sign-up for challenge updates and alerts regarding upcoming webinars, registration details, and data set availability!