DYNAMIC SUMMARY GENERATION FOR INTERPRETABLE MULTIMODAL DEPRESSION DETECTION
Shiyu Teng 1 , Jiaqing Liu 1 , Hao Sun 1,2 , Yu Li 1 , Shurong Chai 1 , Ruibo Hou 1 , Tomoko Tateyama 3 Lanfen Lin 2 , Yen-Wei Chen 1,∗ · 2026
Read DYNAMIC SUMMARY GENERATION FOR INTERPRETABLE MULTIMODAL DEPRESSION DETECTION by Shiyu Teng 1 , Jiaqing Liu 1 , Hao Sun 1,2 , Yu Li 1 , Shurong Chai 1 , Ruibo Hou 1 , Tomoko Tateyama 3 Lanfen Lin 2 , Yen-Wei Chen 1,∗ (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2604-11334.txt.
Description
Teng and eight colleagues detect depression from multiple modalities with a staged framework that pairs language-model summaries with clear reasoning. The pipeline screens for depression, classifies severity into five classes, and regresses a continuous score. At each stage, the language model writes step by step richer clinical notes to guide fusion of text, audio, and video features. All summaries merge into a concise plain report. Tests on E-DAIC and CMDC beat leading baselines on accuracy while staying clear.
Catalog record
| CREATOR | Shiyu Teng 1 , Jiaqing Liu 1 , Hao Sun 1,2 , Yu Li 1 , Shurong Chai 1 , Ruibo Hou 1 , Tomoko Tateyama 3 Lanfen Lin 2 , Yen-Wei Chen 1,∗ |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 14946 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2604.11334