Organizing timely treatment in multi-disciplinary care

A framework for dynamic ICU bed management in a public-private multi-hospital system
Antoine Soiré

ABSTRACT

We consider a dynamic ICU bed management problem and propose a methodology that seeks to identify policies for transferring patients (classified according to diagnosis-related groups) between public hospitals and to private clinics. The methodology is based on the formulation of a discounted infinite-horizon Markov Decision Process and the use of an affine approximation architecture to represent the value function in this model. We adopt the linear programming approach to Approximate Dynamic Programming and solve the corresponding equivalent linear program using column generation. The resulting approximate optimal policies are somewhat counterintuitive as they proactively transfer some patients to other public hospitals and derive others to private clinics, even when beds are still available at the arriving hospital. The performance of the proposed methodology is evaluated by simulating its performance for a practical case based on the problem faced by the hospital network in the city of Santiago, Chile.

BIOGRAPHY

Dr. Antoine Sauré is an Assistant Professor at the Telfer School of Management at the University of Ottawa (Canada). His research interests include stochastic modeling, optimization, and decision-making under uncertainty. He has over twelve years of experience developing, implementing and applying advanced analytics methods to large-scale business problems in several industries. As a member of a multidisciplinary team including researchers, oncologists and key decision makers from the British Columbia Cancer Agency, he worked on the development of numerous capacity planning and patient scheduling systems aimed to provide cancer patients with timely access to quality care.