Tracking Mood and Mood Disorder Using Mobile Sensing

Measuring how people feel is essential to aid in the diagnosis and prevention of emotion dysfunction and disorders such as major depressive disorder. While the experience sampling method (ESM) is the gold standard for capturing emotions in daily life, it has several drawbacks such as high participant burden. On the other hand, mobile sensing, which involves analysing behavioural and contextual data from smartphones, has been heralded as delivering on the promise of providing continuous, objective, and real-time data. However, the potential promise of mobile sensing has not yet been fully fulfilled. Therefore, we propose two principal avenues to help mobile sensing deliver on its promise. First, we will investigate the potential reasons why mobile sensing works well for some groups, but not for others. We will explore whether individuals with more variable emotional patterns may produce more distinguishable behavioural and contextual patterns underlying their emotions, allowing for better prediction of momentary states. Secondly, we will explore alternative uses of mobile sensing as a replacement for standalone, direct predictions of emotion by combining mobile sensing data with ESM for event-contingent triggering of ESM prompts, thereby reducing participant burden in long-term assessments while increasing the emotional information gleaned from the sampled data.

Keywords:Mobile sensing, Affect dynamics, Machine learning, Depression, Multilevel modelling

Disciplines:Statistics and data analysis, Motivation and emotion, Engineering psychology, Psychological assessment, General psychology not elsewhere classified