Exploring digital phenotypes associated with depression and anxiety using smartphone sensors in randomized clinical trials

Abstract

Developing countries such as Brazil struggle to meet mental health demands due to insufficient workforce. Given the economic impacts of the coronavirus crisis, it is likely that this scenario worsens. Therefore, it is important to develop low-cost, scalable strategies to diagnose, deliver, and evaluate the effects of mental health interventions. Smartphones can be useful in this context, since they have embedded sensors that record a significant amount of data, activity logs, and user-generated content. Data generated by these sensors contain patterns that might be robust proxies of behaviors, symptoms, and physiological parameters. The analysis of the relation between signals derived from smartphone sensors and behaviors of interest, named digital phenotyping, has been increasingly explored in the last decade as a method for automatically monitoring mental health symptoms, but its potential to assess changes associated with mental health interventions is yet to be determined. The objective of this project is to implement and test automated and intensive monitoring methods using smartphone sensors to assess symptoms of anxiety and depression in two Randomized Clinical Trials (RCTs) evaluating the efficacy of remote interventions in adolescents (RCT1) and young mothers (RCT2). The specific objectives are: a) to determine the usability and acceptability in relation to digital phenotyping methods; 2) to explore the associations between depression and anxiety symptoms obtained via self-report scales with data derived from smartphone sensors; 3) to verify if associations between sensor data and symptoms of depression and anxiety change throughout time and according to the intervention group; 4) to verify if identified digital phenotypes might predict treatment response. (AU)