Reducing Length of Stay in Vocational Rehabilitation Through Scheduling Policies and Client Classifications

Abstract

This dissertation explores capacity planning in healthcare-based organizations with changing client mix, competency centered training and highly variable length of stay. The data for analysis is provided by a state residential vocational rehabilitation center in the Eastern U.S. The research uses artificial intelligence (AI) and supervised learning in the forms of CART analysis, logistic regression and neural network analysis to categorically classify jobs, patients, clients or customers into groups to better predict length of stay (LOS). Scheduling policies and capacity planning under changing client mix are examined through simulation modeling. The results of this research will provide a mechanism for effective resource use and capacity planning in organizations with highly variable completion times and changing environmental conditions.