JMIR Publications

ABSTRACT

Background:

Depression is highly recurrent in nature and heterogeneous in individual course. The unobtrusive collection of intensive longitudinal passive data through smartphones and smartwatches might enable the mapping of individual courses of depression, differentiating between various illness states and predicting changes in symptoms, thus improving the self-management of symptoms.

Objective:

The goal of this systematic review is to assess whether objectively measured data collected via smartphones and wearable devices can discriminate among different states of depression and whether they can predict changes in mental health states (remission, relapse, or depressive episodes with their severity of symptoms) within a person.

Methods:

We searched PubMed and Web of Science databases for studies published from 2012 to 2022 that (a) included a smartphone app or wearable device; (b) had participants who were diagnosed with depression; (c) had participants who were ≥14 years; (d) had a minimum duration of 12 weeks of continuous data collection; and (e) articles written in English. This systematic review was pre-registered in PROSPERO (Registration number: CRD42022355696).

Results:

Out of 12,997 peer-reviewed articles, we found nine original studies that met the inclusion criteria. The selected studies recruited a range of 45 to 2,200 participants. Study durations ranged from 12 weeks to one year. Three studies derived passive longitudinal data from smartphones of participants, one study used a wearable device to generate data, and five studies used both smartphone and wearable device data. For monitoring purposes, a variety of parameters were collected through sensors by different studies. For example, the most collected variables were number of steps per day, total distance moved per day, smartphone usage, call logs, sleep data, heart rate, light exposure, and speech patterns. Discriminating among different states of depression (worsening, relapse, or recovery) was feasible in one study (1 out of 9, 11.1%). Six studies (6 out of 9, 66.7%) demonstrated predicting capabilities of objectively measured data for depressive episodes and/ or symptom severity. Predictive accuracy for measures of depression based on objectively measured data collected via smartphones and wearable devices was reported in four studies and ranged from 79% to 91%.

Conclusions:

Real-time passive monitoring using wearable devices could be utilized to develop personalized self-management approaches. However, significant gaps still exist in this area, including the lack of longitudinal and long-term studies (e.g., > three months), studies lacking confirmatory parameters on an individual level, and studies without a strong correlation between parameters in individual patients to support clinical decision-making. Improvements in reporting standards are highly recommended to provide better-informed insights for clinicians. Throughout this process, there is a clear need to address various other issues, such as limited types of collected data, reliability, user adherence, and privacy concerns. Clinical Trial: Registration number: CRD42022355696