Open Access

Hybrid Intelligent Model for Mental Health-Oriented Sentiment Mining Across Reddit and Twitter Using Machine Learning and Pretrained Deep Learning Architectures

4 Faculty of Information Technology, Hanoi University of Science and Technology, Hanoi, Vietnam
4 School of Computing, Saigon Institute of Technology, Ho Chi Minh City, Vietnam, Research Interests: Cloud Infrastructure, Distributed Systems

Abstract

The increasing prevalence of mental health discussions on social media platforms such as Reddit and Twitter has created a significant opportunity for computational systems to support early detection of psychological distress through sentiment mining. This study proposes a hybrid intelligent model that integrates classical machine learning techniques with pretrained deep learning architectures to analyze mental health-oriented sentiments across heterogeneous social media sources. The motivation stems from the rising burden of anxiety-related and depressive disorders globally, including conditions such as agoraphobia, panic disorder, and generalized anxiety disorder, which require early identification for effective intervention (Balaram & Marwaha, 2023; Generalized Anxiety Disorder, n.d.).

The proposed framework leverages feature engineering approaches alongside transformer-based contextual embeddings to enhance classification accuracy and domain adaptability. A comparative synthesis of prior research highlights the strengths and limitations of machine learning models, psycholinguistic approaches, and transformer-based architectures in mental health detection tasks. Experimental design considerations include cross-platform dataset alignment, sentiment normalization, and hybrid fusion strategies combining FastText, BERT-like embeddings, and supervised classifiers.

Findings from the literature-driven methodological synthesis suggest that hybrid architectures outperform standalone models in capturing both lexical and contextual emotional cues. However, challenges remain in domain shift, annotation bias, and ethical considerations in mental health prediction systems. The study concludes that hybrid intelligent systems represent a scalable and adaptable direction for mental health sentiment mining across social platforms.

Keywords

References

Agoraphobia. (2023, February 13). Balaram, K., & Marwaha, R. StatPearls - NCBI Bookshelf.
Anxiety disorders. (2023b, September 27). https://www.who.int/news-room/factsheets/detail/anxiety-disorders
Aragon, M., Lopez Monroy, A. P., Gonzalez, L., Losada, D. E., & Montes, M. (2023). DisorBERT: A Double Domain Adaptation Model for Detecting Signs of Mental Disorders in Social Media. Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers).
Bracher-Smith, M., Crawford, K., & Escott-Price, V. (2020). Machine learning for genetic prediction of psychiatric disorders: a systematic review. Molecular Psychiatry, 26(1), 70–79.
Boettcher, N. (2021). Studies of Depression and Anxiety Using Reddit as a Data Source: Scoping Review. JMIR Mental Health, 8(11), e29487.
Depression detection via harvesting social media | Proceedings of the 26th International Joint Conference on Artificial Intelligence. (n.d.). Guide Proceedings.
Feriante, J., Torrico, T. J., & Bernstein, B. (2023, February 26). Separation Anxiety Disorder. StatPearls - NCBI Bookshelf.
Gautam, S., Jain, A., Chaudhary, J., Gautam, M., Gaur, M., & Grover, S. (2024). Concept of mental health and mental well-being, it’s determinants and coping strategies. Indian Journal of Psychiatry, 66(Suppl 2), S231–S244.
Generalized Anxiety Disorder. (n.d.). National Institute of Mental Health (NIMH).
Hannah, K., Marie, K., Olaf, H., Stephan, B., Andreas, D., Michael, L. W., Till, B., & Peter, D. (2023b, November 13). The global economic burden of health anxiety/hypochondriasis-a systematic review. BMC Public Health.
Hussain, J., Satti, F. A., Afzal, M., Khan, W. A., Bilal, H. S. M., Ansaar, M. Z., Ahmad, H. F., Hur, T., Bang, J., Kim, J. I., Park, G. H., Seung, H., & Lee, S. (2019, August 12). Exploring the dominant features of social media for depression detection. Journal of Information Science, 46(6), 739–759.
Ilias, L., Mouzakitis, S., & Askounis, D. (2024, April). Calibration of Transformer Based Models for Identifying Stress and Depression in Social Media. IEEE Transactions on Computational Social Systems, 11(2), 1979–1990.
Jiang, Z., Levitan, S. I., Zomick, J., & Hirschberg, J. (2020, January 1). Detection of Mental Health from Reddit via Deep Contextualized Representations.
Kim, J., Lee, J., Park, E., & Han, J. (2020b). A deep learning model for detecting mental illness from user content on social media. Scientific Reports, 10(1).
Kumar Singh, K. (2023, January 31). Study of Early Risks of Depression by Analysing Social Media Posts. IIMS Journal of Management Science.
Merinda Lestandy, Amrul Faruq, Adhi Nugraha, & Abdurrahim. (2024, April 28). Analyzing Reddit Data: Hybrid Model for Depression Sentiment using FastText Embedding. Jurnal RESTI (Rekayasa Sistem Dan Teknologi Informasi), 8(2), 288–297.
Panic Disorder. (n.d.). National Institute of Mental Health (NIMH).
Social Anxiety Disorder (Social Phobia). (n.d.). Professional, C. C. M. Cleveland Clinic.
Tavchioski, I., Koloski, B., Ε krlj, B., & Pollak, S. (2022). E8-IJS@LT-EDI-ACL2022 -BERT, AutoML and Knowledge-graph backed Detection of Depression. Proceedings of the Second Workshop on Language Technology for Equality, Diversity and Inclusion.
Trifan, A., Antunes, R., Matos, S., & Oliveira, J. L. (2020). Understanding Depression from Psycholinguistic Patterns in Social Media Texts. Lecture Notes in Computer Science, 402–409.
Wong P. (2010). Selective mutism: a review of etiology, comorbidities, and treatment. Psychiatry (Edgmont (Pa. : Township)), 7(3), 23–31.
Zirikly, A., & Dredze, M. (2022). Explaining Models of Mental Health via Clinically Grounded Auxiliary Tasks. Proceedings of the Eighth Workshop on Computational Linguistics and Clinical Psychology.

Similar Articles

1-10 of 63

You may also start an advanced similarity search for this article.