Methods Protocol: Can AI-enabled analyses of smartphones and wearable data collected over one year provide patients suffering from depression with an objective marker of disease severity?

This proof‑of‑concept study aims to analyze whether AI‑based individual analysis of long‑term time series comprising data collected via smartphone and smartwatch (i.e., voice, activity, smartphone usage, self‑ratings) can provide at least some patients with objective information about changes in the severity of their depression. It will thereby address the research question RQI.a as described in (Reich et al., 2025), by answering the following specific research questions: 1. In how many of the enrolled patients with clinically diagnosed depression can analysis of their individual, up to one year passive time series data (variables described in section Data Collection) identify objective markers that estimate the severity of depressive symptomatology as measured by daily PHQ-2 scores? 2. For patients in whom a significant linear prediction is achieved, which sensor or app derived data source contributes most to the explanation of individual depression severity? 3. How does the application of non-linear ensemble methods influence the results of research questions 1 and 2? I.e., how many additional patients exhibit a reliable predictive pattern, and do the dominant data sources shift when non linear relationships are allowed? Reich, H., Schreynemackers, S., Amin, R., Ludwig, S., Zippelius, J., Leimhofer, J., Dunker, T., Schriewer, E., Carell, A., Weber, Y., & Hegerl, U. (2025). Links between self-monitoring data collected through smartphones and smartwatches and the individual disease trajectories of adult patients with depressive disorders: Study protocol of a one-year observational trial. Contemporary Clinical Trials Communications, 45, 101492. https://doi.org/10.1016/j.conctc.2025.101492