An important aspect of this research is the data collection. 100,000 individuals from the UK were asked to wear a movement sensor for seven days during their daily life. The method to detect when individuals are sleeping was recently developed by dr. Vincent van Hees, Senior Research Engineer at Netherlands eScience Center. Sleep detection is not an easy task as inactivity during the evening and in the morning can easily be confused for extended bedtime and vice versa. In the absence of a reliable reference method to train such classifier with machine learning, a heuristic approach based on knowledge about the data was used to classify the data. You can find details of the method, released as open source software, in the following article: " Estimating Sleep Parameters using an Accelerometer without Sleep Diary ".
Dr. van Hees is a co-author of " Genome-wide association analyses of chronotype in 697,828 individuals provides insights into circadian rhythms ". Additional publications using Dr. van Hees' method to gain better insights in sleep and health are expected soon.