Figure 1.

A schematic overview of the PSYCHE-D (Prediction of Severity Change-Depression) model. Phase 1c uses screener survey responses (regarding sociodemographics and chronic comorbidities at baseline), self-reported lifestyle and medication changes (LMC) survey data from the month in which the Patient Health Questionnaire-9 (PHQ-9) label is generated, and data from consumer-grade wearables to categorize each individual’s likely PHQ-9 category. In the second phase, this generated information is combined with the initial PHQ-9 category, screener survey responses, additional LMC self-reports, and consumer-grade wearable device person-generated health data (PGHD) to make the final prediction of whether the individual is likely to have experienced increased depression severity over the 3-month period. Red blocks represent Phase 1, and blue blocks represent Phase 2. C: collected. G: generated.