Prediction of Mental Health Risk Levels for Social Media Users Based on Multi-head Attention Mechanism Optimization of Bidirectional Gated Recurrent Unit Networks

Xu Qihao; Liu Chang; Li Zehui; Jia Yifan; Guo Yifan; Wang Pengkai

doi:10.54254/2753-8818/2026.CH30481

Abstract

In the field of social media mental health monitoring research, machine learning algorithms serve as the core support, establishing a crucial link between multi-source data and mental health status assessment. In view of the inherent flaws of traditional models, this paper constructs a mental health risk prediction model based on the Bidirectional gated Recurrent Unit (BiGRU) network. The research first conducts data distribution analysis and correlation analysis on the dataset, and then introduces multiple machine learning algorithms for classification and comparison. The results show that the proposed model demonstrates a comprehensive leading advantage in all evaluation indicators. Its accuracy rate, recall rate, precision rate and F1 value all reach 95.4%. Not only is it significantly higher than the 82.9%, 82.9%, 83.5%, 82.8% of Naive Bayes, 89.9%, 89.9%, 90.4%, 89.9% of ExtraTrees, and 87.7%, 87.7%, 87.9%, 87.7% of gradient boosting decision trees, It far exceeds support vector machines whose core classification indicators are only in the range of 54.3% to 57.9%, and also has a slight improvement compared to the suboptimal CatBoost. In terms of the AUC metric reflecting the ability to distinguish categories, this model achieved a high score of 99.6%, approaching the perfect classification level. It is slightly better than CatBoost's 99.2% and 99.0%, and significantly outperforms other comparison models, demonstrating superior classification accuracy and category discrimination ability. This model provides an efficient solution for the precise identification of mental health risks on social media, which is of great significance for promoting the practical development of mental health monitoring technology.

Keywords: Multi-head attention mechanism , Bidirectional gated circulation unit , Mental health risk level

Article Information

Received: 4 December 2025

Published: 11 December 2025

DOI: https://doi.org/10.54254/2753-8818/2026.CH30481

Article Type: Research Article

Cite this article

Qihao,X.;Chang,L.;Zehui,L.;Yifan,J.;Yifan,G.;Pengkai,W. (2025). Prediction of Mental Health Risk Levels for Social Media Users Based on Multi-head Attention Mechanism Optimization of Bidirectional Gated Recurrent Unit Networks. Theoretical and Natural Science,151,77-85.

Data availability

The datasets used and/or analyzed during the current study will be available from the authors upon reasonable request.

Disclaimer/Publisher's Note

The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of EWA Publishing and/or the editor(s). EWA Publishing and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

About this Volume

Volume Title: TNS Vol.151

Part of Series: Theoretical and Natural Science

ISSN: 2753-8818 (Print) / 2753-8826 (Online)

Copyright & License

[1]Salvi, Utkarsh , et al. "Decoding Mental Health: Leveraging Machine Learning for Social Media Analysis." #i{Doctoral Symposium on Human Centered Computing} Springer, Singapore, 2025.

[2]Purohit, Shardha , et al. "Analyzing the Impact of Social Media Usage on Mental Health: A Machine Learning Approach." #i{2024 15th International Conference on Computing Communication and Networking Technologies (ICCCNT)} (2024): 1-6.

[3]Azzolina, Danila , et al. "Editorial: Machine learning approaches for monitoring mental health and substance abuse using social media data." #i{Frontiers in Public Health} (2025).

[4]Pachava, Vengalarao , et al. "Machine Learning Analysis of Social Media's Impact on Mental Health of Indian Youth." #i{International Research Journal of Multidisciplinary Scope} 5.2(2024): 623-635.

[5]Kim, Jina. , D. Lee , and E. Park . "Authors' Reply to: Bibliometric Studies and the Discipline of Social Media Mental Health Research Comment on "Machine Learning for Mental Health in Social Media: Bibliometric Study"." #i{Journal of medical Internet research} 23.6(2021): e29549.

[6]Resnik, Philip , et al. "Bibliometric Studies and the Discipline of Social Media Mental Health Research. Comment on "Machine Learning for Mental Health in Social Media: Bibliometric Study" (Preprint)." #i{Journal of medical Internet research} (2021).

[7]Gao, Yan. "Research on the Application of Chinese Language Teaching System based on Artificial Intelligence Technology." 2024 Third International Conference on Distributed Computing and Electrical Circuits and Electronics (ICDCECE). IEEE, 2024.

[8]Sharan, Roneel V. , et al. "Macro-Sleep Staging With ECG-Derived Instantaneous Heart Rate and Respiration Signals and Multi-Input 1-D CNN& #x2013; BiGRU." #i{IEEE Transactions on Instrumentation and Measurement} (2024): 73.

[9]Tang, Yikai , et al. "Degradation Prediction of Proton Exchange Membrane Fuel Cell Based on Multi-Head Attention Neural Network and Transformer Model." #i{Energies (19961073)} 18.12(2025).

[10]Wang, Yuxiao , C. Suo , and Y. Zhao . "Multi-head attention-based variational autoencoders ensemble for remaining useful life prediction of aero-engines." #i{IOP Publishing Ltd} (2024).