Health Outcome Prediction with Multiple Models and Dempster-Shafer Theory

Can multiple predictive models be combined to predict health care outcomes? In this paper, we explore this question by considering the use of multiple predictive models as "evidence" and formulate a multi-model approach to prediction based on Dempster-Shafer's Theory of Evidence. Given the accuracy measures of multiple models, we propose a formulation of a combined predictive model based on Dempster-Shafer's Theory. We then evaluate this approach on a set of data and compare it to predictions by the individual models.