This study demonstrates how Large Language Models (LLMs) can transform prediction of emergency department returns for mental health patients, providing both superior accuracy and clinical interpretability.

• Combined machine learning and LLM approach improved prediction of 30-day ED returns for mental health patients
• Enhanced interpretability makes AI insights more actionable for clinical teams
• Addressed a significant healthcare challenge where 24-27% of mental health patients return to emergency departments within 30 days
• Demonstrated practical application of LLMs in healthcare decision support

This research matters because it bridges the gap between AI capability and clinical utility, potentially reducing the burden on emergency departments while improving mental health patient outcomes through more accurate risk assessment and intervention.

Leveraging Large Language Models to Enhance Machine Learning Interpretability and Predictive Performance: A Case Study on Emergency Department Returns for Mental Health Patients