Dynamic time warping to model daily life stress reactivity in a clinical and non-clinical sample - An ecological...

Dysregulated stress reactivity has been closely linked to psychopathology. However, most studies have applied cross-sectional lab-based methodologies rather than longitudinal designs, like ecological momentary assessment (EMA). EMA provides dynamic, time-sensitive data that require advanced analytical approaches. Dynamic time warping (DTW) is particularly suited for capturing the temporal dynamics between EMA-items.

We applied DTW to examine daily life stress reactivity in 99 participants with a current affective disorder and 277 controls. Using EMA-data, we assessed emotional reactivity to social stressors (i.e., unpleasant or missing company), event-related stressors (i.e., unpleasant experiences), and positive events (i.e., pleasant company or experiences). DTW distances between affect items, stressors, and positive events were calculated to construct group-level networks reflecting undirected and temporal relationships.

The current affective disorder group reported more unpleasant and fewer pleasant events compared to controls, rated their company as pleasant less often, and reported missing company more frequently. In undirected analyses, they showed strong connections between positive and negative emotions, while these formed distinct clusters in controls. Accordingly, temporal network analyses revealed stronger emotional responses to stressors in the affective disorder group. Additionally, unpleasant and pleasant experiences significantly preceded emotional changes in this group but not in controls.

Our findings highlight heightened reactivity to daily events in individuals with a current affective disorder. We demonstrate an application of DTW in psychiatry research and showcase its ability to analyze complex time-series data, infer causality, and create directional networks at the item level for a more in-depth study of stress reactivity.