MedGellan: LLM-Generated Medical Guidance to Support Physicians
This addresses the need for hybrid AI-human systems to reduce errors in critical medical tasks, though it appears incremental as it builds on existing LLM and prompting methods.
The paper tackles the problem of medical decision-making by introducing MedGellan, a lightweight framework that uses an LLM to generate clinical guidance from raw medical records, which improves diagnostic performance with gains in recall and F1 score.
Medical decision-making is a critical task, where errors can result in serious, potentially life-threatening consequences. While full automation remains challenging, hybrid frameworks that combine machine intelligence with human oversight offer a practical alternative. In this paper, we present MedGellan, a lightweight, annotation-free framework that uses a Large Language Model (LLM) to generate clinical guidance from raw medical records, which is then used by a physician to predict diagnoses. MedGellan uses a Bayesian-inspired prompting strategy that respects the temporal order of clinical data. Preliminary experiments show that the guidance generated by the LLM with MedGellan improves diagnostic performance, particularly in recall and $F_1$ score.