- Version
- Download 9
- File Size 385.45 KB
- File Count 1
- Create Date 05/01/2026
- Last Updated 05/01/2026
Dr. NextGen: A Hybrid AI Framework for Multi-Modal Diagnosis, Multilingual Support, and Ethical care in Computational Psychiatry
Prasanna G,
Department of Computer Science and Engineering,
Maharaja Institute of Technology Mysore
Affiliated to Visvesvaraya
Technological University (VTU), Belagavi, Karnataka, India prasanna_cse@mitmysore.in
Prakruthi S,
Department of Computer Science and Engineering,
Maharaja Institute of Technology Mysore
Affiliated to Visvesvaraya
Technological University (VTU), Belagavi, Karnataka, India prakruthis_cse@mitmysore.in
Mukuta Manit D,
Department of Computer Science and Engineering,
Maharaja Institute of Technology Mysore
Affiliated to Visvesvaraya
Technological University (VTU),
Belagavi, Karnataka, India mukutamanitd6@gmail.com
Nihar J,
Department of Computer Science and Engineering,
Maharaja Institute of Technology Mysore
Affiliated to Visvesvaraya
Technological University (VTU), Belagavi, Karnataka, India niharj777@gmail.com
Meghana B,
Department of Computer Science and Engineering,
Maharaja Institute of Technology Mysore
Affiliated to Visvesvaraya
Technological University (VTU), Belagavi, Karnataka, India meghambbs2003@gmail.com
Sandeep K R,
Department of Computer Science and Engineering,
Maharaja Institute of Technology Mysore
Affiliated to Visvesvaraya
Technological University (VTU), Belagavi, Karnataka, India Sandeepkr20044156@gmail.com
A B S T R A C T
The growing prevalence of mental health disorders including Depression, Anxiety Disorder, Schizophrenia, Bipolar Disorder, Obsessive-Compulsive Disorder (OCD), and Post-Traumatic Stress Disorder (PTSD), highlights the need for scalable diagnostic systems supported by strong ethical safeguards. This study presents the (Dr. NextGen) platform, a comprehensive hybrid AI framework designed to enhance connectivity and deliver personalized, multimodal mental healthcare. The system utilizes a dual-path diagnostic architecture: Prediction Pathway I (P1) employs a Random Forest Classifier on structured, scenario-based inputs (83,564 records total), while Prediction Pathway II (P2) leverages a fine-tuned Mistral-7B Large Language Model for contextual affective state analysis. The P1 model demonstrated exceptional internal performance on the test set of 16,713 records, achieving an overall classification accuracy of 0.99904 across seven classes, with perfect Precision, Recall, and F1-Scores. Although statistically compelling, this result mandates rigorous external validation to ensure generalizability and mitigate data leakage risks. The platform is supported by multilingual features (mt5 for 11 Indic languages) and a HIPAA-ready governance framework, including MongoDB encryption and Role-Based Access Control (RBAC). Final clinical deployment is strictly governed by mandated human-in-the-Loop oversight, ensuring that AI augments rather than replaces professional judgment.
Keywords: Mental Health Screening, Random Forest, Large Language Models, Multilingual AI, Ethical AI, Computational Psychiatry
