P2. AI-Assisted Clinical Decision Support in Aesthetic Surgery: Development and Evaluation of a Retrieval-Augmented Generation Model

Abstract

PURPOSE: Aesthetic surgery encompasses a wide range of procedures that require precise decision-making and up-to-date knowledge of evolving techniques and patient preferences. The vast and continuously expanding literature in this field can be overwhelming for practitioners seeking evidence-based guidance. Recent advances in artificial intelligence (AI), particularly in natural language processing, offer potential to assist in complex medical decision-making. Retrieval-Augmented Generation (RAG) models, which combine information retrieval with text generation, show promise in providing context-aware responses in specialized domains. This study aims to develop and evaluate a RAG model integrated with the GPT-4 API. METHODS: We constructed a RAG model using Python, incorporating 6,545 open-access full-text papers from PubMed related to aesthetic surgery. This model was integrated with the GPT-4 API to optimize information retrieval and create an interactive system. The system was tested with various clinically relevant prompts related to aesthetic surgical procedures. RESULTS: The integrated RAG-GPT-4 system was successfully developed and demonstrated the ability to provide detailed, context-specific information on aesthetic surgery techniques. For example, when queried about the optimal approach for rhinoplasty in a patient with thick nasal skin and a deviated septum, the system provided a comprehensive analysis of surgical options, including open versus closed techniques, cartilage grafting, and postoperative care considerations. CONCLUSION: The integration of a RAG model with the GPT-4 API shows significant promise in assisting plastic surgeons with aesthetic surgery decision-making. By providing rapid access to relevant, evidence-based information, this system could enhance clinical decision-making and improve patient outcomes.

Ähnliche Arbeiten