Mitigating Gender Bias in Depression Detection via Counterfactual Inference

2025-12-01

AI Summary

Audio-based depression detection models suffer from gender bias due to epidemiological gender imbalances in training data, leading the model to learn spurious correlations between gender and depression—resulting in over-diagnosis of females and under-diagnosis of males. To address this fairness issue, we propose a causal-graph-based counterfactual debiasing framework that explicitly models and blocks the direct causal path from gender to prediction outcomes, thereby eliminating gender bias at the mechanistic level. Our method integrates counterfactual inference with state-of-the-art acoustic modeling and performs end-to-end debiased optimization on the DAIC-WOZ dataset. Experiments demonstrate that our approach significantly reduces gender disparity—exceeding 40% reduction in both equalized odds (ΔEO) and demographic parity (ΔDP) gaps—while simultaneously improving overall detection accuracy (AUC increases by 2.3%). It outperforms existing debiasing baselines, offering an interpretable and verifiable causal fairness solution for trustworthy medical AI.

Technology Category

• Machine Learning: Ethics, Bias, and Fairness
• Philosophy and Ethics of AI: Bias, Fairness & Equity
• Computer Vision: Bias, Fairness & Privacy

Application Category

• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Social Networks and Social Media: Fairness and bias in social network and social media analysis

Abstract

Audio-based depression detection models have demonstrated promising performance but often suffer from gender bias due to imbalanced training data. Epidemiological statistics show a higher prevalence of depression in females, leading models to learn spurious correlations between gender and depression. Consequently, models tend to over-diagnose female patients while underperforming on male patients, raising significant fairness concerns. To address this, we propose a novel Counterfactual Debiasing Framework grounded in causal inference. We construct a causal graph to model the decision-making process and identify gender bias as the direct causal effect of gender on the prediction. During inference, we employ counterfactual inference to estimate and subtract this direct effect, ensuring the model relies primarily on authentic acoustic pathological features. Extensive experiments on the DAIC-WOZ dataset using two advanced acoustic backbones demonstrate that our framework not only significantly reduces gender bias but also improves overall detection performance compared to existing debiasing strategies.

Problem

• Mitigates gender bias in audio-based depression detection models.
• Addresses over-diagnosis in females and underperformance in males.
• Uses counterfactual inference to remove gender's direct causal effect.

Innovation

• Counterfactual inference to mitigate gender bias
• Causal graph modeling for decision-making process
• Subtracting direct gender effect from predictions