What Can Medical Education Learn From Facebook and Netflix?

Abstract

We had a sobering realization during our first semesterin medical school: The Web sites that society’s future physicians use to socialize (Facebook) and watch television (Netflix) are managed by more sophisticated al-gorithms than the tools we use to learn medicine. These data-driven companies have developed interfaces and algo-rithms to capitalize on the modern “attention economy,” and they measure success through such metrics as “daily active users ” and “time on site. ” They analyze millions of data points on individual and group use (“Big Data”) to develop personalized recommendations, among other tech-niques, that keep users engaged (1). Given our back-grounds in neuroscience and computer science, we decided to ask whether similar methods could be applied to medi-cal education and discuss potential opportunities and bar-riers to these applications. Medical institutions currently rely on self-reported survey data to understand trainee behaviors and perspec-tives. A neutral, data-driven approach would be to track learner statistics, such as the frequency and duration of time devoted to viewing course documents and answering practice questions. For example, we created a Web- and mobile-learning platform (www.osmosis.org)—used by nearly 10 000 medical students who have collectively an-swered questions more than 1.8 million times—that col-lects performance, duration, recency, confidence, and rat-ing metrics. The site allows us to understand study behavior over time and identify areas of potential improve-