JMIR Publications

ABSTRACT

Background:

Depression is a highly prevalent mental disorder, however it remains undiagnosed and untreated in half of cases. Wearable activity trackers collect fine-grained sensor data characterizing behavior and physiology of users, which could be used for timely, unobtrusive, and scalable depression screening.

Objective:

This study examined the predictive ability of digital biomarkers based on behavioral and physiological data from consumer-grade wearables to detect risk of depression in working population.

Methods:

This was a cross-sectional study of 290 healthy working adults. Participants wore Fitbit Charge 2 for two weeks and completed a health survey including screening for depressive symptoms. We extracted a range of known and novel digital biomarkers characterizing physical activity, sleep patterns, and circadian rhythms from wearables using steps, heart rate (HR), energy expenditure, and sleep data. Associations between severity of depressive symptoms and digital biomarkers were examined with Spearman correlation and multiple regression analyses adjusted for potential confounders. Supervised machine learning was used to predict risk of depression (symptoms’ severity and screening status) from digital biomarkers. For performance evaluation we used k-fold cross-validation and obtained accuracy measures from the holdout folds.

Results:

267 participants were included into analysis. 38 (14%) participants had PHQ-9 score equal or above 10 and were identified as depressed. Greater severity of depressive symptoms was significantly associated with greater variation of night-time HR between 2:00–4:00 am and 4:00–6:00 am, lower regularity of weekday circadian rhythms based on steps and HR, and fewer steps-based daily peaks. Our model predicted depression screening outcome with 78% accuracy (sensitivity–82% and specificity–75%) in the contrasted subsample consisting of participants with high and the lowest risk of depression.

Conclusions:

Discovered digital biomarkers from consumer wearables could indicate increased risk of depression in working population, yet current evidence shows limited predictive ability. Combination of these digital biomarkers could discriminate individuals with high risk of depression from individuals with the lowest risk.