A heart rate variability-driven framework for depression screening leveraging emotion-elicited autonomic divergence

Depression manifests significant emotional dysregulation, characterized by heightened sadness susceptibility and attenuated happiness responsiveness in individuals with depression (IWD). This study employs structured emotion induction protocols to analyze physiological response disparities between IWD and healthy controls (HC) across multiple affective states, establishing empirical foundations for optimizing affective computing-based depression screening.

Dual-phase statistical identification was conducted using Mann-Whitney U tests: initially verifying emotion elicitation validity by comparing HRV features between emotional states and resting conditions, subsequently detecting IWD/HC response differences within each emotion. Machine learning frameworks were then constructed leveraging HRV features and intergroup differential response patterns.

Comparative analysis revealed generally consistent directional patterns and response magnitudes across groups for most features, while critical divergences emerged characterized by heightened sadness reactivity in IWD alongside attenuated happiness responsiveness. Implemented models achieved 76.8% accuracy (AUC = 0.772, 95% CI 0.699-0.841) under sadness-specific conditions, outperforming anger/happiness-induced models (≈ 70% accuracy) and substantially surpassing resting-state baselines.

Systematic investigation of HRV-mediated elicitation patterns through discrete emotion induction confirms clinically significant differential responsiveness between groups, empirically validating heightened sadness susceptibility in IWDs.

These findings offer valuable guidance for refining affective computing-based depression screening algorithms, while contributing to the mechanistic understanding of disorder-specific physiological responses to emotional stimuli.