AIED

LLM-based Virtual Standardized Patients with Response Excessiveness Suppression via Direct Preference Optimization for Medical Interview Examinations

Thu Jul 2, 4:10 PM–4:35 PM · North 205

Part of Simulated Learners and Patients for Educational Practice

Synthetic Data & Simulated Learners Generative AI & Large Language Models Conversational AI & Chatbot Tutors Automated Assessment & Scoring

★ Notable speakers

Masaki Uto ★ — Item response theory for automated essay scoring; Bayesian psychometrics; multidimensional IRT; neural AES

Develops LLM-based virtual standardized patients with Direct Preference Optimization to suppress response excessiveness in medical interview exam practice.

Authors

Naoki Shindo, Masaki Uto