Medical Policies in the Context of Primary Prevention for Cardiovascular Disease

Abstract

Access to electronic health records creates an opportunity to build stochastic models that support healthcare providers' decisions to prevent chronic diseases. As the patient's health conditions vary, decision-makers must apply optimal medical policies that learn from patients' health behaviors and consider their needs. In this dissertation, we present new models that address the following key challenges: (1) understanding how the patient demographics influence the disease progression, (2) developing sequential decision-making models under uncertainty that pursue the best health outcomes for individual patients, and (3) developing sequential decision-making models with limited resources to prevent chronic diseases for a population.

We propose operations research methods to develop policies to prevent cardiovascular diseases. We applied our models to longitudinal data for cardiovascular diseases in a large cohort of patients seen in the national Veterans Affairs health system. The contributions of this work include: (1) Developing an EM algorithm to model patient's health progression, (2) creating a simulation framework to test and analyze different treatment guidelines, (3) developing a sequential decision-making model to define cholesterol monitoring policies that maximize societal benefits, and (4) developing an algorithm for identifying and selecting high-risk patients into adherence-improving interventions. Finally, our modeling framework establishes the analytical and theoretical foundation to build stochastic models that address multiple healthcare opportunities for improvement.