진단·변증 스키마 기반 인공지능 표준화 환자 시스템 (CURA) 개발

Abstract

Abstract This study aims to enhance the clinical competency of Korean medicine students by developing a standardized patient system that integrates the unique diagnostic framework of Korean medicine—pattern identification(辨證) and diagnosis(診斷)—with Large Language Models (LLM), Retrieval-Augmented Generation (RAG), and multi-modal technologies. Using CPX data developed by the National Institute for Korean Medicine Development(NIKOM), we constructed a dataset, derived and designed the clinical workflow, and implemented a standardized patient system named CURA (Clinical Understanding & Response Assistant). CURA engages in natural conversations with learners while dynamically presenting visual materials, allowing the inclusion of multidimensional diagnostic data such as patient photographs, thermal images, lung scans, and tongue images. This enables learners to interpret and analyze both visual and linguistic information comprehensively in alignment with Korean medical diagnostic logic, fostering clinical reasoning processes similar to real-world practice. The AI-driven CURA system generates clinical scenarios, interacts with standardized patients, and produces real-time dialogue reports containing patients’ main symptoms and medical histories. Furthermore, it incorporates an automated evaluation module based on CPX scoring criteria, providing real-time scoring and feedback derived from conversational and observational data. Learners can immediately identify missing questions or diagnostic errors, facilitating self-directed learning, while educators can assess students’ clinical performance using consistent criteria, minimizing evaluation bias. The CURA system developed in this study offers a potential solution to the shortage of evaluators and ensures equitable clinical training opportunities for all students. It contributes to improving the quality of Korean medicine clinical education and addressing educational resource imbalances. Moreover, by providing diverse simulated clinical scenarios, CURA supports competency-based education for Korean medicine students and offers valuable insights into the future direction of Korean medical education.