Harnessing Electronic Health Record Data to Identify Candidates for HIV PrEP

Preexposure prophylaxis (PrEP) can prevent HIV infection, but it is woefully underutilized in part because clinicians don't always identify people at risk in the course of routine clinical encounters. Given that PrEP is one of the pillars of the U.S. plan to eliminate the HIV epidemic, we need to develop new approaches to expand its utilization. Now, two groups have assessed whether applying machine-learning algorithms to data contained in the electronic health record (EHR) can identify individuals at high risk for acquiring HIV.

Marcus and colleagues evaluated EHR data on approximately 3.75 million members of Kaiser Permanente Northern California. The researchers developed a model to predict new HIV infections using 81 risk variables, inclu…