Multimodal Depression Recognition via Mutual Information Maximization Joint With Multi-Task Learning
IEEE Transactions on Biomedical Engineering · 2026-01 · 73 (1) · 369-379
DOI: 10.1109/tbme.2025.3582806 ↗Kurz gefasst
Fachartikel in der Zeitschrift IEEE Transactions on Biomedical Engineering, 2026-01. Verfasst von Yanjie Liu und 7 weiteren Autor:innen. Hauptthemen laut MEDLINE: Depression, Machine Learning. Die Arbeit wurde bisher 1-mal zitiert (Crossref) und nennt 57 Literaturangaben. Gefördert von: Postdoctoral Fellowship Program of the China Postdoctoral Science Foundation, CPSF, National Natural Science Foundation of China.
Themen: Depression
Abstract (Originalsprache)
Depression is a serious mental health disorder with a potential hazard for individuals and society characterized by persistent sadness and hopelessness. Multimodal information including vision, audio, and text is critical for depression diagnosis and treatment. Most studies focus on designing sophisticated feature extraction methods but ignore feature enhancement and fusion within intra-modality and cross-modality. In this paper, a Chinese Multimodal Depression Corpus (CMD-Corpus) dataset is established assisted by clinical experts aiming to support more depression research. Furthermore, we propose a multimodal depression recognition framework based on Mutual Information Maximization with Multi-task Learning (MIMML) to enhance feature representation and fusion among video, audio, and text modalities. The MIMML employs the strategy of maximizing mutual information to accelerate modality-invariance enhancement. The multi-task is used to improve the representation performance of the single modality to improve modality-specific enhancement. Meanwhile, a gated structure with bidirectional gated recurrent units and convolutional neural networks is designed to achieve multimodal feature fusion, which is key to boosting completeness among modalities. Experimental results show that the proposed MIMML effectively captures representation to increase depression recognition accuracy, achieving 84% and 89% accuracy on DAIC-WOZ and our self-collected CMD-Corpus dataset respectively.
Schlagwörter
MeSH-Schlagwörter (MEDLINE)
Förderung
- Postdoctoral Fellowship Program of the China Postdoctoral Science Foundation
- CPSF
- National Natural Science Foundation of China
- Interdisciplinary Research Program of HUST
- Department of Education of Guangdong Province
- Basic and Applied Basic Research Foundation of Guangdong Province
- King Saud University for funding this work
- Research Group
Zitieren
APA
Liu, Y., Liu, Y., Ma, X., Wang, R., Yang, Q., Mo, Y., Alqahtani, S. A., & Chen, M. (2026). Multimodal Depression Recognition via Mutual Information Maximization Joint With Multi-Task Learning. IEEE Transactions on Biomedical Engineering, 73(1), 369-379. https://doi.org/10.1109/tbme.2025.3582806
Vancouver
Liu Y, Liu Y, Ma X, Wang R, Yang Q, Mo Y, et al. Multimodal Depression Recognition via Mutual Information Maximization Joint With Multi-Task Learning. IEEE Transactions on Biomedical Engineering. 2026;73(1):369-379. doi:10.1109/tbme.2025.3582806
BibTeX
@article{Liu2026,
author = {Liu, Yanjie and Liu, Yulan and Ma, Xinran and Wang, Rui and Yang, Qin and Mo, Yijun and Alqahtani, Salman A and Chen, Min},
title = {Multimodal Depression Recognition via Mutual Information Maximization Joint With Multi-Task Learning},
journal = {IEEE Transactions on Biomedical Engineering},
year = {2026},
volume = {73},
number = {1},
pages = {369--379},
doi = {10.1109/tbme.2025.3582806}
}
EUVIMED