ChatENT: Augmented Large Language Model for Expert Knowledge Retrieval in Otolaryngology - Head and Neck Surgery

Abstract Background The recent surge in popularity of Large Language Models (LLMs), such as ChatGPT, has showcased their proficiency in medical examinations and potential contributions to medical education. However, LLMs possess inherent limitations, including inconsistent accuracy, specific prompting requirements, and the risk of generating harmful hallucinations. A domain-specific, fine-tuned model would address these limitations effectively. Methods OHNS-relevant data was systematically gathered from open-access internet sources and indexed into a database. We leveraged Retrieval-Augmented Language Modeling (RALM) to recall this information and used it for pre-training, which was then integrated into ChatGPT 4·0, creating a OHNS specific knowledge Q&A platform known as ChatENT. Findings ChatENT showed enhanced performance in the analysis and interpretation of OHNS information, outperforming ChatGPT 4.0 in both the Canadian Royal College OHNS sample examination questions challenge and the US board practice questions challenge, with a 58.4% and 26·0% error reduction, respectively. ChatENT generated fewer hallucinations and demonstrated greater consistency. Interpretation To the best of our knowledge, ChatENT is the first specialty-specific LLM in the medical field. It appears to have considerable promise in areas such as medical education, patient education, and clinical decision support. The fine-tuned model has demonstrated the capacity to overcome the limitations of existing LLMs, thereby signaling a future of more precise, safe, and user-friendly applications in the realm of OHNS. Funding The authors received no financial support for the research, authorship, and/or publication of this project.