PhD Thesis Proposal: Vincent Chen

ZOOM LINK
Meeting ID: 931 7655 7929
Passcode: 789237

"Artificial Intelligence for Fair Decision-Making in Healthcare"

Abstract

Artificial intelligence has significant potential to improve healthcare decision-making, but its use in clinical and patient-facing settings also raises important concerns about fairness, reliability, and patient-centered outcomes. This thesis studies AI methods for fair decision-making in healthcare, with a focus on developing models that improve decision quality while accounting for heterogeneity across patient populations.

The thesis consists of three main chapters. The first chapter studies machine learning and optimization methods for fair decision-making in healthcare, focusing on how predictive models and decision rules can be designed to balance utility and fairness. The second chapter develops Boosted Distributional Reinforcement Learning, a framework for analyzing treatment policies and healthcare decision processes through the full distribution of long-term outcomes rather than only their expectations. This approach is motivated by healthcare applications where both average benefit and outcome variability across patients are important. The third chapter studies a fair multi-armed bandit framework for large language model-based systems with patient feedback in the loop. This chapter explores how adaptive decision-making methods can incorporate feedback over time while encouraging fair and reliable recommendations across patient groups.

Together, these projects aim to build AI systems that are not only predictive and adaptive, but also fair, interpretable, and better aligned with healthcare decision-making needs.

Thesis Committee

• Wesley Marrero (Chair)
• Eugene Santos, Peter Chin
• Shuhang Wang (Staff Data Scientist, Walmart)

Contact

For more information, contact Thayer Registrar at thayer.registrar@dartmouth.edu .