Towards Accurate Non-Intrusive Recollection of Stress Levels Using Mobile Sensing and Contextual Recall

Existing user input sampling methods used for stress-related psychotherapy are plagued with problems of low recall accuracy and high intrusiveness. In this paper, we propose a contextual recall-based self-report method that uses mobile sensing technology to capture contextual cues including location, activity, and environmental acoustics to aid accurate recollection of stress levels associated with stressful events. We conducted a controlled user study with 36 participants. Our experimental results suggest that contextual recall outperforms recall-based self-report method by increasing the recall accuracy and minimizing intrusiveness to participants at the same time. Moreover, we quantified the contribution of each individual contextual cue in recollecting stress levels. We found that although participants perceived all the contextual cues to be useful, in reality, not all the contextual cues are weighted equally during the recollection process.