Update: Jan 2023, This work now is accepted at JMIR Formative Research.
Feeling proud and grateful to represent Mindstrong at APS 2022 (Association of Psychological Science) this year. Trained as a quantitative psychologist on intensive longitudinal data analysis from Penn State, it’s long been my desire to connect the dots between mental health and passive sensing.
In this digital phenotyping work (aka APS poster), we (Xiao Yang, Jonathan Knights, Audrey Klein, Victoria Bangieva, Holly DuBois, Justin Baker) provide an empirical analysis of 145 members at Mindstrong regarding their behavior gleaned from passive sensing and their depressive symptoms over a year’s time.
First, we created interpretable behavioral features that are expected to be related to depression based on theory and previous literature. Then we utilized the nested structure of longitudinal data to run multilevel modeling and delineated the within- and between-person association. Third, we found several significant behavioral features that are associated with depressive symptoms (spoiler alert: app usage and typing behavior showed promising results, see more details in the poster). This analysis supports the utility of passively collected digital behavior to assess depressive symptoms.
We are only scratching the surface concerning the human-phone interaction behavior and its potential to continuously monitor depression (see a snippet of the data in the poster). There is a lot more to discover in this interdisciplinary space, e.g., clinical psychology, data science, and behavior sensing, just to name a few. More exciting work to be done!