Can you identify a pill by its shape, color, and markings? Chances are you can’t, but you know of times when it would be useful. You could use your smartphone to take a picture of a mystery pill and almost immediately see what it most likely could be. The National Library of Medicine (NLM), part of the National Institutes of Health, would like all of us to be able to more easily identify pills. And it is working on a fix, starting with prescription pills. NLM recently launched its Pill Image Recognition Challenge (Federal Register Notice, Submission Instructions). The Challenge asks any and all innovators to design algorithms and software that can effectively match lower-quality photos – like those we take with smartphones – with high-quality publicly-available images in the NLM RxIMAGE database. Today’s RxIMAGE has images and descriptions of about 4,000 prescription pills. NLM hopes that the satisfaction of addressing this easily stated problem, along with monetary awards, will entice experts and students working in fields as diverse as image recognition and data analytics to participate. NLM researchers plan to use Challenge submissions and potentially work with Challenge participants to build a software system and an Application Programming Interface (API) underlying a consumer-facing smartphone app for image recognition of prescription pills. The NLM Pill Image Recognition Project team thinks that people who pick up refills of prescriptions for generic drugs for themselves or family members would benefit from using such an app. A refill can be made by a different manufacturer and can have different color, shape, or markings. That can be confusing, but the app could help confirm what the refill is for. We look forward to seeing how the Pill Image Recognition Challenge can leverage the RxIMAGE database to produce tools for such public benefit.
HHS Competes, an HHS IDEA Lab program, supports the use of prizes, challenges, and crowdsourcing to more effectively leverage the intelligence of the crowd to solve our nation’s toughest problems. Learn more.