Hilltop Institute study on predictive healthcare models

Predictive models have proliferated in the health system in recent years and have been used to predict both health services utilization and medical outcomes. Less is known, however, on how these models function and how they might adapt to different contexts. 

The paper, authored by Ruichen Sun, Morgan Henderson, Leigh Goetschius, Fei Han, and Ian Stockwell, reports on the results of a study on the inner workings of a large-scale predictive model deployed in two distinct populations, with a particular emphasis on adaptability issues. 

They found that the model adapted to, and performed well in, both populations despite demographic differences in these two groups. However, the most salient risk factors and their relative weightings differed, sometimes dramatically, across the two populations. The unadapted Medicaid model displayed poor performance relative to the adapted model.