Changes in smartphone-sensed behaviours predict changes in depression symptoms

Changes in sensor-derived behavioural features using a smartphone are associated with subsequent changes in mental health symptoms among people with elevated symptoms of depression, but there appears to be no evidence that depression changes correlate with subsequent changes in sensed behaviours, according to a study.

“This supports a bidirectional relationship in which changes in movement patterns precede symptom change, but symptom change does not precede changes in movement behaviours,” the researchers said.

This longitudinal study assessed continuously collected phone sensor data and symptom severity data, obtained every 3 weeks over 16 weeks, in 282 adult participants (mean 38.9 years, 79.1 percent female, 80.1 percent White) who owned Android smartphones recruited through national research registries.

The researchers classified participants into four groups: multiple comorbidities, depression and generalized anxiety, depression and social anxiety, and minimal symptoms. Primary outcomes were depression (8-item Patient Health Questionnaire), generalized anxiety (Generalized Anxiety Disorder 7-item scale), and social anxiety (Social Phobia inventory) severity.

Among participants with multiple comorbidities, changes in GPS features (time: r, –0.23; p=0.02; locations: r, –0.36; p<0.001), exercise duration (r, 0.39; p=0.03), and use of active apps (r, –0.31; p<0.001) preceded changes in depression. [J Med Internet Res 2021;23:e22844]

“GPS location features were fairly consistently and negatively, albeit modestly, related to subsequent changes in depression severity across symptom groups,” the researchers said. “This is consistent with a number of previous relatively small studies showing correlations between GPS features and depression.” [J Med Internet Res 2015;17:e175; Peer J 2016;4:e2537; IEEE Trans Biomed Eng 2017;64:1761-1771; Depress Anxiety 2019;36:72-81; Psychiatr Rehabil J 2015;38:218-226]

In depression and anxiety groups, changes in GPS features for locations (r, –0.20; p=0.03) and transitions (r, –0.21; p=0.03) preceded changes in depression.

Of note, depression changes did not correlate with subsequent sensor-derived features, and the minimal symptoms group showed no significant associations. Additionally, sensor-based features showed no relationships with anxiety but had minimal associations with social anxiety.

Among clusters characterized by depression and social anxiety, increases in telephone calls correlated with increases in social anxiety symptoms. The capacity for sensor-derived features to be associated with changes in social anxiety symptoms was not as consistent as that for depression symptoms.

“These findings indicate that sensor-derived behavioural features, which are objective and can be acquired with reduced participant effort, can be a useful tool for investigating the role of some behaviours in changing depressive symptom severity,” the researchers said.

Previous studies had speculated about the clinical potential of mobile sensing. The effect sizes were modest but consistent with several studies that had examined the use of sensed behavioural features to estimate the presence or severity of symptoms. [J Med Internet Res 2017;19:e62; J Am Med Assoc 2017;318:1215-1216; Curr Psychiatry Rep 2015;17:602]

“This study does not support the use of phone sensor data alone in monitoring symptom changes in mental health populations; however, phone sensor data may be useful in conjunction with other networked sensing tools such as wearables,” the researchers said.

“Phone sensor data may be useful for digital mental health interventions,” they added. [NPJ Digit Med 2019;2:88]