Stochastic Control and Optimization Methods for Chronic Disease Monitoring and Control, Hospital Staffing, and Surgery Scheduling.

Abstract

Two critical challenges facing healthcare providers, and more broadly society, are controlling the cost of providing care to patients and improving the health outcomes. This dissertation leverages operations research and analytics to develop patient-centered decision support tools to improve the quality of care, patient safety, and timeliness of access to care at lower cost. This will result in bending the cost curve of healthcare spending and can contribute to a positive impact on health outcomes and improve the quality of life for many people. Three specific research topics are treated. The first research topic focuses on integrating and personalizing the monitoring and control of chronic diseases to improve the quality of care and avoid unnecessary testing and treatment. Our modeling paradigm incorporates each patient's past and present readings in a feedback-driven control model to optimally determine (1) when to schedule office visits and which subset of test to perform to monitor for disease progression (exploration), and (2) what levels of controllable disease risk factors should be targeted to slow the rate of disease progression (exploitation). Glaucoma is discussed as a case study. The second research topic strives to reduce medical errors at hospitals by reducing the number of error-prone patient handoffs via designing a new work shift schedule for healthcare providers. Previous research has focused on the communication aspects of handoffs and has provided recommendations to achieve high-quality handoffs. While a high-quality, structured handoff process is important, decreasing number of patient handoffs is an additional and fundamental way to reduce opportunities for medical errors caused by communication breakdowns, thereby supporting safer and more efficient patient care. The third topic of this dissertation is focused on providing timely access to care, which not only affects patient satisfaction but also directly influences patient safety and health outcomes. We present novel ideas on how to creatively use patient information to coordinate clinic and surgery appointment scheduling. The decision support tool proactively books a clinic and a tentative surgery appointment for all patients with different acuity levels to ensure access to surgery within a maximum wait time target that is clinically safe for them.