Although several studies have identified strong heterogeneity in comorbidity profiles of patients readmitted to the hospital, current CMS models do not use such information. Here we used bipartite networks to automatically identify biclusters of patients and comorbidities, and used them to develop stratified predictive models, with implications for precision medicine.

Learning Objective: 1. Learn why stratified predictive models are important for analyzing hospital readmission
2. Understand how bipartite networks can be used to develop stratified predictive models.

Authors:

Suresh Bhavnani (Presenter)
UTMB

Clark Andersen, University of Texas Medical Branch
Yu-Li Lin, University of Texas Medical Branch
Emmanuel Santillana, UTMB
Tianlong Chen, UTMB
Yong-Fang Kuo, University of Texas Medical Branch

Presentation Materials:

• Abstract/Manuscript - 1