Abstract :

Mental Health Disorders Such As Depression, Anxiety, Bipolar Disorder, And Attention Deficit Hyperactivity Disorder (ADHD) Affect Millions Of People Worldwide And Require Timely Identification For Effective Intervention. Traditional Diagnostic Approaches Often Depend On Clinical Evaluations, Which May Not Provide Continuous Monitoring Of An Individuals Psychological Condition. With The Increasing Use Of Social Media, Large Volumes Of User-generated Text, Images, And Audio Content Offer Valuable Insights Into Emotional And Behavioral States. This Research Proposes TriModalSenseNet, A Novel Multimodal Deep Learning Framework For Early Mental Health Disorder Detection Using Social Media Data. The Framework Integrates Textual, Visual, And Audio Modalities To Capture Comprehensive Behavioral Characteristics. Text Features Are Extracted Using BERT, Image Features Through ResNet-50, And Audio Representations Using Wav2vec 2.0. An Attention-Based Deep Neural Network Performs Multimodal Feature Fusion, While A Long Short-Term Memory (LSTM) Network Models Temporal Behavioral Changes Over Time. Experimental Results Demonstrate The Effectiveness Of The Proposed Approach In Classifying ADHD, Anxiety, Bipolar Disorder, And Depression. The Model Achieved 92.60% Accuracy, 91.60% Precision, 91.20% Recall, And 91.00% F1-score, With The Best Performance Obtained At Epoch 20. The Findings Indicate That Multimodal Learning Combined With Temporal Analysis Significantly Enhances Mental Health Prediction Performance, Supporting Intelligent Healthcare Systems For Early Detection And Intervention. Keywords: Mental Health Detection, TriModalSenseNet, Multimodal Learning, BERT, ResNet-50, Wav2vec 2.0, Attention Network, LSTM, Social Media Analytics.

Published:

21-9-2026

Issue:

Vol. 26 No. 9 (2026)

Page Nos:

310 - 319

Section:

Articles

License:

This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.

How to Cite