Our laboratory published an article in the international journal Big Data and Cognitive Computing (Scopus, Q1):

Ryumina E., Axyonov A., Dolgushin M., Ryumin D., Karpov A. DEPART: Multi-Task Interpretable Depression and Parkinson's Disease Detection from In-the-Wild Video Data // Big Data and Cognitive Computing, 2026, vol. 10(3), 89. (WOS IF=5.3 Q1, Scopus SJR=1.033 Q1)

Automated detection of depression and Parkinson's disease from video data can contribute to the development of scalable and non-invasive health monitoring systems. However, existing methods often focus on a single disease and provide limited interpretability, while different conditions may co-occur in real-world data. The paper presents a novel interpretable multi-task approach, DEPART (DEpression & PArkinson's Recognition Technique), for simultaneous detection of depression and Parkinson's disease from videos recorded in uncontrolled conditions. The approach includes body-region extraction, visual feature encoding using CLIP, Transformer-based temporal modeling, and prototype-aware classification with a gated fusion mechanism. Gradient-based attention maps are used to interpret model decisions by visualizing the image regions that are most relevant to each task. Experiments are conducted on the In-the-Wild Speech Medical (WSM) corpus. The multi-task model achieves Recall=82.39% for depression detection and Recall=78.20% for Parkinson's disease detection. Error analysis reveals the impact of annotation characteristics, static visual content, and occasional body-detection failures. After cleaning the test data, the multi-task model achieves Recall=87.50% for depression and Recall=86.14% for Parkinson's disease. The results demonstrate the potential of joint and interpretable analysis of multiple diseases using video data recorded in real-world conditions.