Essays on the management of appointments for chronic conditions

摘要

Treating chronic conditions is a fairly complex task, which requires well-timed appointments to control one's disease progression. In my dissertation I would like to optimize the monitoring strategies and better predict the demand-for-care of patients with chronic kidney disease (CKD). To do that I design a chronic disease monitoring framework which consists of forecasting, survival analysis and Markov Decision Process (MDP) models. First, I propose a forecasting model which quantifies the impact of CKD-related doctor's appointments on patient's disease progression. The model accounts for patient's comorbidities, vital signs, and important laboratory values. Second, I propose a survival analysis model, which estimates the expected life days of a patient given his or her current health status. Finally, I use the information gained from the first two models to parametrize and solve the MDP, which can suggest monitoring strategies and predict medium-term demand for CKD-patient-care in a clinic. In addition to the chronic disease monitoring framework, I examine CKD patient characteristics associated with a higher resource utilization.