Predictive models have proliferated in the health system in recent years and have been used to predict both health service utilization and medical outcomes. Less is known, however, on how these models function and how they might adapt to different contexts.
In a new article in Medical Care titled Behind the Curtain: Comparing Predictive Models Performance in 2 Publicly Insured Populations, Hilltop researchers Morgan Henderson, Leigh Goetschius, Fei Han, and Ruichen Sun—along with UMBC researcher Ian Stockwell—share the findings of a study they conducted on the inner workings of the Hilltop Pre-AH Model™, a large-scale predictive model that predicts the risk of avoidable hospitalizations and has been deployed in two distinct populations—Medicare and Medicaid—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. Moreover, applying the Medicare model risk factor weights to the Medicaid model—in a so-called unadapted model—displayed poor performance relative to the adapted Medicaid model.
Read more about Hilltop’s work in predictive modeling.
For all media inquiries, please contact Marsha Willis.
|
