The Potential of Large Language Models in Treating Recurrent Medical Conditions through Enhanced History.pdf

This paper presents a method for improving diagnostic accuracy using a <b>Large Language Model (LLM)</b> integrated with a <b>Bayesian probabilistic framework</b>. The approach involves iteratively refining diagnostic probabilities through structured questioning, where each question is designed to maximize information gain and reduce uncertainty. By updating the likelihood of potential diagnoses with each new piece of evidence, the system demonstrates significant improvement in diagnostic precision. The results show that this method outperforms traditional diagnostic techniques, achieving a <b>140.5% improvement</b> in accuracy, highlighting the potential of AI-assisted decision-making in clinical settings.