Title: Human Expertise and Adaptation in Healthcare Operations: From Diagnostic Discrepancies to Generative AI

Abstract: How does professional expertise translate into operational performance when clinical information evolves and generative AI enters service delivery? This seminar brings together two empirical studies examining expertise and adaptation in healthcare. The first study examines diagnostic discrepancies, defined as mismatches between admitting and principal diagnoses, as retrospective indicators of operational burden rather than direct measures of diagnostic error. Using 5.7 million emergency-department-origin inpatient stays in Florida, we find that discrepancy-flagged stays have 6.62% higher hospitalization costs after extensive patient-risk adjustment and high-dimensional fixed effects. This differential is smaller for physicians with greater condition-specific experience but larger for those with broader experience across diagnostic categories. Exploratory analyses reveal corresponding differences in diagnostic and procedural intensity and time to first procedure. These findings distinguish the roles of depth and breadth in professional expertise and inform retrospective monitoring and case-review prioritization. The second study investigates how human counselors respond when a generative AI agent enters a mental health platform as a parallel service provider. A matched difference-in-differences design across two platforms finds increased counselor engagement and an estimated 11.9% increase in paid counseling appointments. LLM-based content analysis and counselor interviews provide evidence consistent with social comparison, whereby visible AI responses and patient endorsement help counselors assess the agent’s capabilities and recalibrate their own effort and communication. Counselors selectively converge toward the agent’s informational-emotional content profile, while interviews also reveal efforts to preserve distinctive relational strengths. Together, the studies highlight experience-task alignment and professional adaptation as complementary perspectives on healthcare operations, with implications for operational monitoring, workforce learning, and the design of services involving both human and AI providers.

Short Bio: Seokjun Youn is an Associate Professor of Management Information Systems and a Levine Faculty Fellow at the Eller College of Management, University of Arizona. His research lies at the intersection of healthcare operations, digital platforms, and AI-enabled decision-making to study how professionals and technologies shape service performance. His work combines large-scale observational data with econometrics, causal inference, optimization, and machine learning methods. His work appears in Manufacturing & Service Operations Management, Production and Operations Management, Journal of Operations Management, and Information Systems Research, and received the 2026 POMS College of Healthcare Operations Management Best Paper Award, the 2024 Decision Sciences Institute Best Theory-Driven Empirical Research Paper Award, and other honors. He serves as an Associate Editor for Decision Support Systems and on the editorial review boards of Production and Operations Management and Decision Sciences Journal. He holds a Ph.D. in Operations and Supply Chain Management from Mays Business School at Texas A&M University, an M.S. in Industrial Engineering from Texas A&M University, and a B.S. in Industrial Engineering from Seoul National University. Before academia, he served as an officer in the Republic of Korea Air Force.