Fig. 3
Metrics of data quality for the dataset. Within each panel, the x-axis shows each unique patient-day sorted and grouped by patient, the y-axis shows a specific metric of data quality. Colors represent different individuals, with a legend in the bottom right panel. Within each panel, patients are arranged by descending mean total data coverage, defined in “Methods”. For a–d (first column, or accelerometer) and e–h (second column, or GPS), each panel from top to bottom gives the estimated number of bursts per day, the daily average frequency of pings per burst, the daily average duration per burst (in seconds), and the daily average duration or gap between bursts (in seconds), respectively. The black lines show the expected values for these measurements, which are defined in the “Methods” section. i–l (third column, or surveys) from top to bottom show the time between subjects first responding to each phone survey after receiving a prompt (in seconds), the time from first response to survey completion (in seconds), and the total number of surveys taken per day, respectively
