MultiMindNet: AI-based mental health analysis using hybrid deep learning approach and Hybrid Ant-Grey Wolf
Sunil Kumar Sharma1,2, Ahmad Raza Khan3, Ghanshyam G Tejani4,5
1Department of Information Systems, College of Computer and Information Sciences, Majmaah University, Majmaah, Saudi Arabia.
PLOS Digital Health
|April 13, 2026
Summary
This study introduces NeuroHAGWO-Net, an AI framework using electroencephalogram (EEG) and text data for accurate mental health detection. The model achieves high accuracy, offering a robust tool for early screening and clinical support.
Area of Science:
- Neuroscience
- Artificial Intelligence
- Computational Psychiatry
Background:
- Mental health disorders present significant global challenges, necessitating accurate and non-invasive diagnostic methods.
- Current diagnostic approaches rely on subjective and time-consuming self-reports or clinical evaluations.
- There is a critical need for automated, objective tools for early mental health screening.
Purpose of the Study:
- To propose NeuroHAGWO-Net, an advanced AI framework for automated mental health status detection using multimodal data.
- To integrate electroencephalogram (EEG) signals and behavioral textual data for enhanced diagnostic accuracy.
- To develop a robust system for early and reliable mental health screening.
Main Methods:
- Utilized Empirical Mode Decomposition (EMD) for EEG signal noise reduction.
- Employed Bidirectional Encoder Representations from Transformers (BERT) for behavioral text data embedding.
- Developed a hybrid BiLSTM-CNN architecture for EEG analysis and integrated behavioral embeddings for multimodal fusion.
- Implemented a novel Hybrid Ant-Grey Wolf Optimization (HAGWO) for feature selection.
- Performed AI-based detection using NeuroVisionNet, combining EfficientNetV2 and Temporal CNNs (T-CNNs).
Main Results:
- Achieved high performance metrics on behavioral and EEG datasets.
- Demonstrated accuracy of 0.9945, precision of 0.9874, and sensitivity of 0.9935 on behavioral data.
- Reported excellent overall performance, including F1-Score of 0.9909 and MCC of 0.9925.
- Showcased low False Positive Rate (FPR) of 0.0151 and False Negative Rate (FNR) of 0.0092.
Conclusions:
- NeuroHAGWO-Net is a highly accurate and efficient AI framework for mental health detection.
- The multimodal approach integrating EEG and text data enhances screening capabilities.
- The model offers a robust solution for early mental health screening and clinical support, leveraging advanced optimization and deep learning.