Predictive modeling of health trajectories: preventable hospitalizations in diabetic mellitus

Project Description

This PhD project (which is part of a larger program of work in changing health trajectories) aims to ‘understand and measure’ risk in existing populations. The ‘risk’ in this case refers to the probability that subjects who are diagnosed with diabetes will have a subsequent hospitalization, deemed preventable, for diabetes or a related condition. We loosely define a hospitalization to be preventable if a patient following physician instructions with respect to disease care would be unlikely to encounter it.

Project Objectives

The project has two specific objectives:

1. Understand and measure risk in existing populations. Before starting to identify which consumers face what type of risks, one needs to be able to define accurately, understand, and measure those risks using available claims data.
2. Develop individual-level predictive models for risk. Models may be specific to one type of risk (say, diabetes), but the methodology needs to be generalizable to other types.