Comparison of bedtime rest classification from accelerometer placed on waist or wrist calculated using Sadeh's algorithm and the decision tree with classification obtained using whole- room indirect calorimeter.

a- Area under the ROC curve calculated as sensitivity multiplied by specificity before data was rounded;b- defined as the probability of correctly classifying bedtime rest period;c- defined as a probability of correctly classifying activity period;d- Wilcoxon signed rank test;e- automated computer algorithm;f- cut points were 20 counts/min (bedtime) and 500 counts/min (activity);g- cut points were 250 counts/min (bedtime) and 3000 counts/min (activity).