Seminar Abstract: Healthcare decisions are rarely made with full information: risks evolve, service durations fluctuate, and data are often incomplete. Overlooking such uncertainties in the optimization process can compromise safety, access, and efficiency. In this seminar, I will examine key healthcare decision-making problems, such as ambulatory care appointment scheduling and hospital readmission reduction, each characterized by different types of uncertainty (observable arrivals, partially observable health status, and complex-to-model unpunctuality). I will present cutting-edge solution approaches (e.g., approximate dynamic programming, predict-and-optimize) that translate these uncertainties into actionable decisions. I will also demonstrate their computational efficiency, performance relative to state-of-the-art benchmarks, and potential impact in real-world healthcare settings.

Presenter Bio: Amirhossein Moosavi is a Michigan Data Science Fellow at the University of Michigan with a PhD in Management Science and MS/BS in Industrial Engineering. He blends machine learning and advanced analytics to enable robust decisions in healthcare and generate evidence that informs policy. At the University of Michigan, Amirhossein serves as Postdoctoral Affairs Chair, and mentors undergraduate and graduate students. Off campus, he serves as one of five national INFORMS representatives and is co-launching an international forum for doctoral and postdoctoral scholars to share their work and expand their networks. His long-term vision is to fuse novel data-driven methods with efficient healthcare delivery to generate insights that inform policymaking. Outside academia, he enjoys strength training, cycling, and kayaking.