Predicting Risk of Re-hospitalization for Congestive Heart Failure Patients

Thesis (Master's)--University of Washington, 2012

Congestive Heart Failure (CHF) is one of the leading causes of hospitalization, and studies show that many of these admissions are readmissions. Identifying patients who are at a greater risk of hospitalization, can guide implementation of appropriate plans to prevent these readmissions. In the field of medical sciences, prediction of such outcomes is a challenging task since it involves integration of various variables associated with patients, such as patients' socio-demographic factors, health conditions, health care utilization and factors related to health care providers. This work aims at analyzing the problem and building an effective predictive model to identify patients who are at a greater risk of future hospital admissions. We propose several classification algorithms to that end. The precursory step to the actual model building process is the information extraction phase; this step seems to be prohibitively challenging due to the prevalence of noise in the dataset, heterogeneity and diverse nature of the sources, and sparsity to name a few. Our initial results are encouraging, as we significantly outperform the existing predictive model proposed by the researchers at Yale University. Our solutions are empirically evaluated by using a health care data set provided by Multicare Health System (MHS).