Topic Modeling Based Multi-modal Depression Detection

TL;DR

This paper introduces a novel topic modeling approach for multi-modal depression detection that effectively captures temporal information in long interviews, outperforming existing methods.

Contribution

The paper proposes a new topic modeling based method for context-aware analysis of multi-modal data in depression detection, addressing temporal information loss.

Findings

Outperforms baseline methods on all metrics

Effectively captures temporal details in long interviews

Improves depression level prediction accuracy

Abstract

Major depressive disorder is a common mental disorder that affects almost 7% of the adult U.S. population. The 2017 Audio/Visual Emotion Challenge (AVEC) asks participants to build a model to predict depression levels based on the audio, video, and text of an interview ranging between 7-33 minutes. Since averaging features over the entire interview will lose most temporal information, how to discover, capture, and preserve useful temporal details for such a long interview are significant challenges. Therefore, we propose a novel topic modeling based approach to perform context-aware analysis of the recording. Our experiments show that the proposed approach outperforms context-unaware methods and the challenge baselines for all metrics.