[1Win4-79] Proposed Method for AI Patient Modeling Using Large Language Models
Keywords:Medical Education, Simulated Patients, Large Language Models (LLMs)
The medical interview constitutes a cornerstone of diagnostic processes for healthcare professionals and represents a critical skill for fostering trust with patients and effectively obtaining pertinent information. However, teaching this skill necessitates the involvement of simulated patients—individuals specifically trained to portray patient roles—whose availability is often limited in rural areas compared to urban centers. To address this limitation, the present study fine-tuned a large language model (LLM), specifically GPT-4o-mini, utilizing conversation data from medical interviews to develop an AI-based simulated patient capable of generating contextually appropriate responses. This study leveraged the LLM's inherent conversational generation capabilities within predefined disease scenarios while addressing key challenges associated with LLMs, such as hallucinations and excessively verbose responses.
Authentication for paper PDF access
A password is required to view paper PDFs. If you are a registered participant, please log on the site from Participant Log In.
You could view the PDF with entering the PDF viewing password bellow.