The predictive tool analyzed 44 variables within the medical records of millions of HIV-negative patients at Kaiser Permanente Northern California. The model flagged 2.2 percent of those patients as having high or very high risk of contracting HIV; after three years, those flagged patients comprised nearly 40 percent of the members of the test group who had been diagnosed with HIV.
The model could therefore identify many more patients who could benefit the most from pre-exposure prophylaxis (PrEP), which is currently used by just 7 percent of those whom it could benefit, per CDC estimates. Healthcare providers and previous models have difficulty flagging patients with high HIV risk since they typically focus only on factors such as sexual orientation and history of sexually transmitted infections; the new, AI-powered model improves upon these predictions by assessing dozens more factors.
The study was led by researchers from Kaiser Permanente San Francisco and the health system’s division of research, as well as Harvard Medical School and Beth Israel Deaconess Medical Center, both in Boston.
More articles about AI:
In healthcare, ethical AI is a life-or-death issue: Q&A with AI Ethics Lab’s founder and director
AI develops ‘turbocharged’ flu vaccine
12 AI initiatives launched by hospitals, health systems in 2019
