Inter-individual differences in sleep quality: Insights from mining wearable sleep-tracking data

Many people are using mobile and wearable technologies to collect their sleep data in order to learn more about their sleep patterns and to identify sleep anomalies. However, these sleep data are rarely analyzed in a systematic way and many users find it difficult to understand if their sleep patterns are normal. To address this issue, we applied a data mining technique to analyze personal sleep tracking data collected from 7 participants, aiming to investigate inter-individual difference in terms of abnormal sleep patterns. We found that population criterion may classify people who are satisfied with their sleep experience as poor sleepers. On the other hand, sleep anomalies are missed for people whose personal sleep distribution deviates from sleep distribution of the population. These abnormal sleep patterns manifested themselves in different dimensions. For example, one person may have problems in sleep duration, while another person may sleep enough but have lots of awakenings. Even within one dimension, abnormal sleep patterns of one person can be antithetical to those of another person. The quantitative analysis results on personal sleep data support the conclusions of previous qualitative studies on sleep tracking, suggesting the need for a change in evaluating sleep quality that takes into account interpersonal differences.