Abstract:
|
Health disparities are a key area of interest at the regional, and national levels. While there is literature on the understanding of disparities at the population level, there is limited understanding of factors at the hospital level, which are building blocks for the population level trends. A dearth of statistical methods, which can distill out factors at the hospital levels are among the key contributors to this lack of understanding. In this proposal, innovative learning methods are proposed which can lead to better understanding of such factors. Novelties in the proposal include sophisticated adjustments for factors such as rare events; a wholistic and nuanced treatment of ethnicity and race data beyond naïve techniques; and, multilevel risk adjustment calculators, which can contribute to efficient models and improved inference.
|