Skip to main content

Table 3 Comparison of algorithm, visual inspectio n, and logs in the identification of the window of consecutive days of wear

From: Development and application of an automated algorithm to identify a window of consecutive days of accelerometer wear for large-scale studies

  Total sample n = 169 Sample with logs n = 74
Visual inspection vs. algorithm N (%) Visual inspection vs. logs N (%) Algorithm vs. log N (%)
Same number of days in wear window
 Complete agreement 93 (55.0) 50 (67.6) 39 (52.7)
 Wear window shifted by 1 day in one method 46 (27.2) 10 (13.5) 15 (20.3)
 Average difference in hours of wear 0.3 (more in visual inspection) 5.3 (more in visual inspection) 1.5 (more in algorithm)
 Wear window shifted by ≥2 days in one method 2 (1.2) 0 (0) 1 (1.4)
 Average difference in hours of wear 2.2 (more in algorithm) N/A 4.5 (more in algorithm)
Total, N (%) 141 (83.4) 60 (81.1) 55 (74.3)
Different number of days in wear window
 Differed by 1 day 25 (14.8) 13 (17.6) 19 (25.7)
Method with more days of wear Visual inspection Visual inspection See footnotea
Differed by ≥2 days 3 (1.8) 1 (1.4) 0 (0)
Method with more days of wear Visual inspection Visual inspection N/A
Total, N (%) 28 (16.6) 14 (18.9) 19 (25.7)
  1. aAlgorithm had more days for 9 (12.2%) signals; log had more days for 10 (13.5%) signals.
\