AdoDAS: A Privacy-Preserving Multimodal Challenge for Adolescent Depression, Anxiety, and Stress Assessment

Zhaojie Luo, Haizhou Li, Hiroshi Ishiguro, Tetsuya Takiguchi, Tomoko Matsui, Kun Qian, Fei Wang, Shuqiong Wu, Zhengjun Yue, Junkun Wang, Tianhua Qi

Keywords: Multimodal Signal Processing, Adolescent Depression, Anxiety, Stress Assessment

TL;DR: AdoDAS is a large-scale, privacy-preserving multimodal dataset of 6,000 youths and 24,000 audio–video segments, with feature-level data, DASS-21 labels, and standardized benchmarks for adolescent depression, anxiety, and stress assessment.

Abstract: We present AdoDAS, a privacy-preserving multimodal grand challenge dataset for adolescent Depression/Anxiety/Stress (D/A/S) assessment. AdoDAS contains 6,000 child and adolescent participants and 24,000 audio–video segments collected via a controlled school-based protocol that combines a standardized reading passage with open-ended interview prompts. Ground-truth labels are derived from the DASS-21, providing three subscale scores (D/A/S) and 21 item-level responses to support both coarse screening and fine-grained, interpretable symptom modeling. Unlike recent grand challenges that focus on social-media text or adult interview settings, AdoDAS targets minors and addresses stringent privacy constraints by withholding raw recordings and releasing reproducible pre-computed representations and temporal metadata. We provide two benchmark tracks with strong baselines to facilitate robust, subject-disjoint evaluation and advance safe multimodal mental-health research.

Email Sharing: We authorize the sharing of all author emails with Program Chairs.

Data Release: We authorize the release of our submission and author names to the public in the event of acceptance.

Submission Number: 15