Skip to main page content
U.S. flag

An official website of the United States government

Dot gov

The .gov means it’s official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2025 Aug 16;8(1):523.
doi: 10.1038/s41746-025-01933-3.

AI-assisted multi-modal information for the screening of depression: a systematic review and meta-analysis

Luyao Wang #  1 , Chenhan Wang #  1 , Chenyang Li  1 , Toshiya Murai  2 , Yicai Bai  1   3 , Ziyan Song  1 , Shuoyan Zhang  1 , Qi Zhang  1 , Yu Huang  1 , Xiaoying Bi  4 , Jiehui Jiang  5   6
Affiliations

AI-assisted multi-modal information for the screening of depression: a systematic review and meta-analysis

Luyao Wang et al. NPJ Digit Med. .

Abstract

Depression is a prevalent and costly mental disorder across all ages. Artificial intelligence (AI)-assisted physiological and behavioral information-such as electroencephalography (EEG), eye movement, video or audio monitoring, and gait analysis-offers a promising tool for depression screening. We systematically reviewed the classification performance of these AI-assisted measures in depression screening. A comprehensive literature search was conducted in Google Scholar, Web of Science, and IEEE Xplore, with the search date up to June 7, 2025. The reported AUC values are pooled estimates calculated from all results of eligible studies. AI-assisted multi-modal methods achieved a pooled AUC of 0.95 (95% CI: 0.92-0.96), outperforming uni-modal methods (pooled AUC: 0.84-0.92). Subgroup analysis indicated deep learning models showed higher performance, with an AUC of 0.95 (95% CI: 0.93-0.97). These findings highlight the potential of AI-based multi-modal information in depression screening and emphasize the need to establish standardized databases and improve research design.

PubMed Disclaimer

Conflict of interest statement

Competing interests: The authors declare no competing interests.

Figures

Fig. 1
Fig. 1. Brain activation patterns in depression across key functional domains.
Brain activation patterns in depression across key functional domains, including altered emotional processing, negative attentional bias, delayed auditory responses, and impaired sensorimotor-emotional integration. BG Basal Ganglia, ATL Anterior Temporal Lobe, FFA Fusiform Face Area, VP Ventral Pallidum, SC Superior Colliculus, OC Occipital Cortex.
Fig. 2
Fig. 2. PRISMA flowchart of study selection.
Illustration of the identification, screening, eligibility assessment, and inclusion of studies in the systematic review.
Fig. 3
Fig. 3. Hierarchical SROC curves of studies in meta-analysis of depression and NC classification using different paradigms.
a The SROC curves comparing uni-modal and multi-modal approaches. b The SROC curves stratified by deep learning and machine learning methods.
Fig. 4
Fig. 4. Outlook for multi-modal physiological and behavioral information integration.
Illustration of Multi-modal Fusion, AI Model Updating, Dataset Standardization, and Cross-group Research in Diverse Applications.
See this image and copyright information in PMC

Similar articles

See all similar articles

References

    1. Diseases, G. B. D. & Injuries, C. Global burden of 369 diseases and injuries in 204 countries and territories, 1990-2019: a systematic analysis for the Global Burden of Disease Study 2019. Lancet396, 1204−1222 (2020). - PMC - PubMed
    1. Mitchell, A. J., Vaze, A. & Rao, S. Clinical diagnosis of depression in primary care: a meta-analysis. Lancet374, 609–619 (2009). - PubMed
    1. de Aguiar Neto, F. S. & Rosa, J. L. G. Depression biomarkers using non-invasive EEG: A review. Neurosci. Biobehav. Rev.105, 83–93 (2019). - PubMed
    1. Stolicyn, A., Steele, J. D. & Series, P. Prediction of depression symptoms in individual subjects with face and eye movement tracking. Psychol. Med.52, 1784–1792 (2022). - PubMed
    1. Toto, E., Tlachac, M. & Rundensteiner, E. A. AudiBERT: a deep transfer learning multimodal classification framework for depression screening. In Proceedings 30th ACM International Conference on Information & Knowledge Management 4145–4154. 10.1145/3459637.3481895 (2021).

LinkOut - more resources