Demystifying Mental Health Reports Through an LLM-based Approach

Medical diagnoses typically contain strong jargon that is inaccessible to people with lower health literacy.Mental health is one such domain where comprehensibility is exacerbated further.Even with increasing awareness about mental health, support from practitioners or authentic knowledge sources is required to understand diagnostic information.Through this work, we utilize locally trained and finetuned Large Language Models (LLMs) to examine comprehension improvement of mental health diagnostic reports.Through a survey (n=105) with participants over 18 and a demonstration to practicing professionals, we examine the outputs of our AI-driven approach.Our approach offers two key benefits: (1) it simplifies complex psychiatric information for laypersons, providing understandable explanations; (2) it enables doctors to access concise summaries and information via tailored retrieval techniques.We outline this tool's development, explain the methodology, evaluate its potential for information gain and healthcare delivery, and propose directions for future research.