Developing Explainable AI Models for Personalized Treatment Recommendations Using LLMs and EHR Data

Personalized treatment recommendations remain critical for better outcomes and optimization of healthcare delivery. This paper discusses an explainable AI framework for personalized treatment recommendation using large language models in conjunction with EHR data. Using the publicly available dataset MIMIC-III, this approach integrates into a unique analysis advanced natural language processing capabilities of LLMs, interpreting complex clinical narratives along with structured data. We implement explainability techniques that provide insight into the decision-making process of the AI model to ensure transparency and trust. The results show improved accuracy and clinician satisfaction compared to traditional models, underlining the potential of explainable AI in personalized medicine. This work emphasizes the need to combine large datasets with state-of-the-art AI technologies to enable trustworthy and effective healthcare solutions.