Browse Prior Art Database

Comparing group e ects in risk of adverse healthcare events using Bayesian Hierarchical Model

IP.com Disclosure Number: IPCOM000244978D
Publication Date: 2016-Feb-04
Document File: 6 page(s) / 118K

Publishing Venue

The IP.com Prior Art Database

Abstract

Bayesian Hierarchical Models (BHM) are used for the analysis and comparison of risks associated with interactions among various groups (e.g. patients with different demographic characteristics, healthcare providers, etc.) so that better care management can be provided with the objective of serious risk mitigation. As typical with Bayesian models, the models under consideration provide robust predictions when sample size is small and there are random effects associated with interaction among different group/factors.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 26% of the total text.

Page 01 of 6

 Comparing group effects in risk of adverse healthcare events using Bayesian Hierarchical Model

Lina Fu, Faming Li, Jing Zhou, Xuejin Wen, Jinhui Yao and Michael Shepherd

Abstract

  During the recent few years, the United States healthcare industry is under unprecedented pressure to improve outcome and reduce cost. Many healthcare organizations are leveraging healthcare analytics, especially predictive analytics in moving towards these goals and bringing better value to the patients. While many existing adverse event prediction mod- els provide helpful predictions in terms of accuracy, their use are typically limited to prioritizing individual patients for care management. Health- care organizations are also highly interested in understanding the driving risk factors for these adverse events for the population they are serving, so that they could design system level interventions or policy changes to improve the health outcome for the population broadly. One sensible ap- proach is to compare different entities, such as hospitals, ethnicity groups and geographical regions, in a statistically risk-adjusted fashion, so that they could scrutinize the worse performing entities whie identifying best practices among the better performing ones. In this invention, we de- velop a framework based on Bayesian Hierarchical Model that will allow us to directly compare the effects of different groups, including but not limited to hospitals, so that the healthcare organizations could draw a lot more insights to guide their system-level interventions. We illustrate the approach in the context of hospital readmission modeling.

1 Introduction

The current healthcare system in United States is unsustainable due to the ever rising cost. Particularly the cost in United States is nearly twice as high as in most other developed countries[1]. The rapid cost growth is driven by aging population, the pervasiveness of chronic diseases, the shortage of primary care physician and nursing, inefficiency and lack of quality. Healthcare providers and payers are also under increasing pressure from consumers to improve quality and deliver better outcomes. Therefore, many healthcare organizations are starting to develop models that can predict future adverse healthcare events, such as hospital readmission and emergency department visits, so that they are able

1


Page 02 of 6

to prioritize and deploy individual interventions for higher risk patients. In the past few years we have seen that a variaty of models have provided useful prediction in most cases in terms of accuracy.

  However, as helpful as individual risk prediction could be for planning and deploying individual interventions to reduce risk of future adverse events, these interventions typically require healthcare providers such as doctors, nurses and care managers in the loop and are difficult to scale. Therefore, it is equally important, if not more, for healthcare organizations to understand the driving risk f...