DynaBridge: AI Framework for DASS-Structured Mental Health Assessment
A novel AI framework named DynaBridge has been introduced to evaluate depression, anxiety, and stress through a comprehensive behavioral analysis. Distinct from standard fusion models, DynaBridge utilizes the psychometric framework of the DASS-21 questionnaire, which assesses risk based on a structured arrangement of symptom items through fixed mappings. This framework integrates acoustic, visual, and textual information gathered over several sessions, enhanced by DASS-aware summaries generated by frozen LLMs as participant-specific semantic evidence. It forecasts ordinal item distributions, reconstructs risk evidence from soft scores at the item level, and merges this with direct multimodal predictions. A confidence-aware refinement approach carefully integrates high-confidence semantic indicators. Validation of the framework was conducted using the official AdoDAS dataset, as detailed in arXiv:2607.25679.
Key facts
- DynaBridge is a dynamic summary-guided cross-task multimodal framework for DASS-structured mental health assessment.
- It encodes acoustic, visual, and textual cues across multiple sessions.
- It uses frozen-LLM-generated DASS-aware summaries as participant-level semantic evidence.
- It predicts ordinal item distributions and reconstructs depression, anxiety, and stress risk evidence from item-level soft scores.
- A confidence-aware refinement strategy incorporates high-confidence semantic cues conservatively.
- The framework was validated on the official AdoDAS dataset.
- The research was published on arXiv with ID 2607.25679.
- DynaBridge addresses the psychometric structure of DASS-21 questionnaire labels.
Entities
Institutions
- arXiv