NEW STATISTICAL METHODS TO BETTER LEVERAGE EMERGING HEALTH CARE UTILIZATION DATA

Loading...
Thumbnail Image

Files

Publication or External Link

Date

2020

Citation

Abstract

Improving the healthcare system is an important task that is always both socially and individually beneficial, and statistics is one of the useful tools that have been applied in pursuit of this goal. However, limitations on current methods and the introduction of new forms of data have created many new challenges and research opportunities. It is therefore crucial to explore and extend statistical methods to better understand and leverage healthcare utilization data, particularly recent and emerging data in new forms. In this dissertation, we develop and apply various innovative statistical methods to address five specific healthcare issues. First, we successfully develop a novel approach to model the length of hospital stay using mixture distributions through an EM algorithm. Second, we extend a two-state continuous time Markov chain to estimate patient readmission risk at a large academic hospital in the U.S. Third, we study changes in accessibility in emergency departments from 2016 to 2018 among 21 hospitals in Maryland Region III. Fourth, we investigate the impact of the global budget payment model on emergency department accessibility. Lastly, we use a multi-state Markov model to explore cascading events during emergency room crowding, also in Region III of Maryland.

Notes

Rights