| 研究実績の概要 |
A self-tracking experiment was conducted with a cohort of 27 participants to collect data using three devices concurrently: a Fitbit Charge 2, a Neuroon eye mask, and a medical device. Each participant tracked their sleep for 3 nights in their homes and afterwards was invited for a 1-hour interview which was audio-recorded and transcribed. The study produced rich qualitative data and quantitative data. On the one hand, we analyzed the qualitative interview data to examined people’s mental models of sleep and sleep tracking devices. We found that people generally had incomplete and faulty knowledge on sleep, and there were two blind spots in people’s mental models of sleep tracking devices: not knowing what data was collected and not knowing how raw data was processed. Based on the analysis, we proposed several design recommendations to guide future design of better sleep tracking technologies. On the other hand, we analyzed the quantitative sleep data measured by consumer sleep tracking devices against those derived by the medical device. We found that consumer sleep tracking devices agreed reasonably well to the medical device in terms of total sleep time and sleep efficiency. However, both devices performed poorly in measuring sleep stages. The validation study and the users study revealed that Fitbit Charge 2 is a promising wearable device for accurate home sleep tracking. Using the Generalized Linear Mixed Modelling (GLMM), we designed a new algorithm that achieved better sensitivity and balanced accuracy in comparison with the proprietary algorithm of Fitbit.
|
