MeDiSumQA is a novel dataset and framework that uses large language models to transform complex medical discharge letters into patient-friendly question-answer pairs, addressing the gap between medical documentation and patient health literacy.

• Created from MIMIC-IV discharge summaries to generate accessible medical information
• Enables evaluation of LLMs for safe and patient-centered medical text generation
• Bridges the communication gap between healthcare providers and patients with varying health literacy levels

This research matters because it provides a standardized approach to evaluate AI systems that make medical information more accessible to patients, potentially improving treatment adherence and patient outcomes.

MeDiSumQA: Patient-Oriented Question-Answer Generation from Discharge Letters