Can Conversational Temporal Dynamics Improve Depression Detection in Dyads? A Preliminary Investigation in Multi-Modality Perspectives

2026-07-04

AI Summary

This study addresses the common oversight in existing automatic depression detection methods of the temporal dynamics inherent in clinician–patient interactions during clinical interviews. To this end, it introduces, for the first time, a lightweight dialogue temporal feature (24-dimensional) as an independent and efficient modality. The approach integrates this temporal module with frozen WavLM-large and RoBERTa-large models through an interpretable convex-weight late fusion strategy. The temporal features alone achieve the best single-modality performance on the development set. Upon fusion, the model attains macro F1 scores of 0.804 and 0.669 on the development and test sets, respectively. Notably, the acoustic modality is automatically assigned zero weight in the fusion process, underscoring the efficacy and dominance of the proposed temporal features.

Abstract

Automatic depression detection from clinical interviews typically models the semantic content and acoustic characteristics of participant speech. However, the interactional timing between the clinician and participant remains comparatively under-modeled. We investigate conversational temporal dynamics, specifically dyadic turn-pair timing, as a primary modality fused with self-supervised encoders. Evaluated on the DAIC-WOZ dataset, we compare a compact 24-dimensional timing module against frozen WavLM-large and RoBERTa-large baseline detectors. This temporal module achieves the highest single-modality performance on the development set. Furthermore, a convex-weighted late fusion strategy improves overall performance to 0.804 and 0.669 macro-F1 on the development and test sets, respectively. The learned fusion effectively assigns zero weight to acoustics, demonstrating that conversational timing serves as a lightweight, interpretable complement for dyadic depression screening.

Problem

• depression detection
• conversational temporal dynamics
• dyadic interaction
• turn-pair timing
• multi-modality

Innovation

• conversational temporal dynamics
• dyadic turn-pair timing
• late fusion
• depression detection
• self-supervised encoders