MindWellQA: A Semantically Enriched Evidence-Based QA System for Psychological Disorders
Year: 2025
Type: article
Abstract: The increasing prevalence of psychological disorders highlights the urgent need for early diagnosis and intervention.While advanced humanoid Large Language Models (LLMs) offer potential for intelligent Question-Answering (QA) systems in this domain, existing generic LLMs lack specificity and fail to provide evidence-based responses. Additionally, the lack of context can lead to ambiguous and vague answers, which may result in inaccuracies or a lack of emotional sensitivity.. To address these limitations, we present MindWellQA, an evidence-based, semantically enriched QA system designed to answer queries related to psychological disorders like Mood Disorder, Paranoid Personality Disorder,Anxiety Disorders and Psychotic Disorders including symptoms and treatments. This system integrates the latest scientific inventories, dictionaries, medicine, and therapy databases while leveraging intent classification, named entity recognition, and prompt engineering for enhanced response accuracy. MindWellQA is available in two variants Alpha and Beta based on the complexity of the LLM used in context generation. MindWellQA is built using a transfer learning approach with Parameter-Efficient Fine-Tuning (PEFT) and is trained on a dataset over 2,000 classified questions with different domains like symptoms,treatments and disorders. These questions are augmented with variations based on gender, age, and profession to ensure inclusivity and personalization. The system’s responses are evaluated using both quantitative and qualitative metrics, including word count, TF-IDF, and Rouge variants, ensuring high-quality outputs. Furthermore, MindWellQA undergoes rigorous validation by psychological experts and end-users from diverse demographics. The system demonstrates high accuracy, sensitivity, and user-friendliness, making it a promising tool for individuals seeking reliable psychological health information.As compared to most prevalent LLM/GPT models, an evidence-based approach enhances trust, usability, and accuracy, setting a higher standard for psychological disorder QA systems.
Source: IEEE Access
Authors: Dipti Pawar, Shraddha Phansalkar
Institution: MIT Art, Design and Technology University
Language: English
Subfield: Health Informatics
Field: Medicine
Domain: Health Sciences
Open Access status: gold
Estimated APC paid: $2,075
OpenAlex ID: W4415748046