DIGITAL BIOMARKERS OF DEPRESSION AND ANXIETY: A SYSTEMATIC REVIEW OF SLEEP, ACTIVITY AND SMARTPHONE-DERIVED MEASURES

Background. Depression and anxiety are the primary drivers of disability, and are primarily currently measured through episodic self-report. Smartphones and wearables track sleep, physical activity, mobility and device use, and continuously produce “digital biomarkers” that could be used to monitor objectively and in real-time.

Objective. To integrate published literature and findings regarding the associations and/or predictions of depression and anxiety symptoms with sleep, activity and smartphone measures.

Methods. The systematic review was conducted in line with the PRISMA 2020 (Page et al., 2021) and only published studies were included; no participants were recruited and no questionnaires or interventions were given. The literature identification, screening, extraction and appraisal and synthesis was conducted between January and July 2026. The final evidence base contained a total of 35 sources which were analysis narratively and descriptively, with an adapted QUADAS-2 framework (Whiting et al., 2011) used to appraise the quality of the studies.

Results. There were 15 empirical digital-biomarker studies (43%), 9 reviews and frameworks (26%), 6 meta-analyses or epidemiological syntheses (17%) and 5 measurement or methods standards (14%). Sleep was the most consistent finding across data for depression, physical activity occurrence and irregularity was moderate, and physiological signals and anxiety were the least consistent. Insomnia increased the risk of developing depression by approximately two-fold (pooled OR 2.10), while more physical activity was protective (pooled OR 0.83 and 0.74). Although most of the sensing studies were small and short, and based on students' samples or convenience samples, few studies were validated with external tests.

Conclusions. Signals of sleep, activity and mobility are potential indicators of depressive symptoms, while there is less mature evidence for signals of anxiety. There is a need for bigger, longer, externally validated and clinically diverse trials with the same feature definitions to support clinical decisions with digital biomarkers.