Talk: Patient Prioritization Model With Limited Resources And Stochastic Compliance.

Abstract

Physicians seek to prevent chronic diseases by tracking the patients’ healthcare behavior. They handle a large heterogeneous panel of patients, which is costly or demands multiple resources. Unfortunately, physicians do not count on infinite resources, for which they need to prioritize the panel. Not prioritize patients may forgo needed treatment and suffer adverse events related to the disease. Additionally, despite the physician’s efforts, prioritize patients may not adhere to medication and follow-up recommendations. We present a Multi-armed bandit model to maximize the panel’s total life years gain. Further, we tested the model using longitudinal data for cardiovascular diseases in a large cohort of patients seen in the national Veterans Affairs health system. Finally, we study the resulting prioritization policies and their structure.