Hierarchical attention interpretation: an interpretable speech-level transformer for bi-modal depression detection

TL;DR

This paper introduces a hierarchical attention-based transformer model for bi-modal depression detection from speech, enhancing interpretability and clinical relevance by providing detailed speech and sentence-level explanations.

Contribution

It proposes a novel bi-modal speech-level transformer that avoids segment-level labelling and offers hierarchical interpretability through gradient-weighted attention maps.

Findings

Model outperforms segment-level models with higher accuracy and F1 score.

Provides detailed sentence and token-level explanations for depression detection.

Enables clinicians to verify model predictions with interpretable insights.

Abstract

Depression is a common mental disorder. Automatic depression detection tools using speech, enabled by machine learning, help early screening of depression. This paper addresses two limitations that may hinder the clinical implementations of such tools: noise resulting from segment-level labelling and a lack of model interpretability. We propose a bi-modal speech-level transformer to avoid segment-level labelling and introduce a hierarchical interpretation approach to provide both speech-level and sentence-level interpretations, based on gradient-weighted attention maps derived from all attention layers to track interactions between input features. We show that the proposed model outperforms a model that learns at a segment level ($p$=0.854, $r$=0.947, $F1$=0.897 compared to $p$=0.732, $r$=0.808, $F1$=0.768). For model interpretation, using one true positive sample, we show which sentences within a given speech are most relevant to depression detection; and which text tokens and Mel-spectrogram regions within these sentences are most relevant to depression detection. These interpretations allow clinicians to verify the validity of predictions made by depression detection tools, promoting their clinical implementations.