סמינר המחלקה להנדסת תעשייה

~~Multi-Session Appointment Scheduling with Heterogeneous Clients
Reut Noham
post-doctoral fellow at the Department of Industrial Engineering and Management Sciences at Northwestern University

Abstract:
Clients seeking paramedical therapies and rehabilitation services generally attend frequent appointments over an extended period. Motivated by an early intervention program that provides therapeutic services to infants and toddlers with developmental delays and disabilities, we study scheduling policies that are designed to meet the needs of heterogeneous clients and the operational considerations of the providers. The clients can be heterogeneous in many dimensions: availability and preferences over time, length of service needed, and urgency of need. We aim to better understand how the different ways a provider may prioritize these factors influence scheduling decisions.
The early childhood years present a critical time window in which brain plasticity intensify children's ability to learn new skills. Studies have shown that for a wide range of conditions known to adversely affect developmental progress, such as Cerebral palsy, Down syndrome, or Autism, early intervention makes greater improvements than interventions at later age. Our work is grounded in a partnership with a non-profit organization that provides early intervention services in the Chicagoland area. The non-profit organization matches clients to providers to best meet the clients' needs and the availability of providers.  In this talk, I present our analysis of the single provider problem and discuss extensions to the multi-provider setting.
The problem of assigning clients to available days and time slots of the service provider is described as a Markov Decision Process. Clients are assigned sequentially, and only probabilistic knowledge of future clients is known. Given the characteristic of the client and the availability of the provider, our model determines which client requests (specifying day and slot) the service provider should accept in line with a specified prioritization of the provider. We characterize the structural properties of optimal scheduling decisions under idealized conditions. We then use these properties to develop a heuristic for general cases. We evaluate the performance of this heuristic relative to intuitive rule-of-thumb heuristics. Ultimately, we show that designing dynamic scheduling policies balances the many considerations involved in these scheduling decisions, specifically decreasing the number of rejected requests and improving health outcomes while maintaining high service providers' utilization.
*Joint work with Dr. Karen Smilowitz
 
Bio:
Reut Noham is a post-doctoral fellow at the Department of Industrial Engineering and Management Sciences at Northwestern University. She received her Ph.D. degree in Industrial Engineering from Tel-Aviv University in 2019. Her research interests include supply chain management and logistics with a focus on humanitarian supply chains, healthcare systems, and non-profit optimization. In her current research, she focuses on dynamic models for solving complex operational problems in collaboration with practitioners. Her research is supported by the Tel Aviv -Northwestern Post-Doctoral Fellowship, and the Eric and Wendy Schmidt Postdoctoral Award for Women in Mathematical and Computing Sciences.