Table 2. Medians for the area under curve (AUC), sensitivity, and specificity for various cut points (counts/min) tested in the development sample set using Receiver Operating Characteristic (ROC) curves for accelerometer worn a waist or wrist during a ∼24-h stay in a whole-room indirect calorimeter.

Waist					Wrist
CP1 (counts/min)	CP2 (counts/min)	AUC ± SDa	Sensitivityb	Specificityc	(CP1) (counts/min)	CP2 (counts/min)	AUC ± SDa	Sensitivityb	Specificityc
10	400	0.832±0.174	0.902	0.962	150	3500	0.898±0.078	0.969	0.965
10	500	0.847±0.175	0.922	0.962	175	2500	0.915±0.084	0.961	0.983
10	600	0.847±0.160	0.922	0.962	175	3000	0.922±0.077	0.976	0.983
15	400	0.859±0.149	0.928	0.959	175	3500	0.909±0.072	0.981	0.964
15	500	0.859±0.152	0.958	0.959	200	2500	0.920±0.080	0.976	0.983
15	600	0.860±0.148	0.958	0.954	200	3000	0.929±0.076	0.982	0.982
20	400	0.856±0.150	0.947	0.946	200	3500	0.928±0.071	0.991	0.964
20	500	0.872±0.154	0.983	0.946	225	2500	0.924±0.069	0.981	0.982
20	600	0.870±0.149	0.985	0.933	225	3000	0.929±0.065	0.997	0.980
25	400	0.852±0.150	0.961	0.924	225	3500	0.928±0.061	0.998	0.964
25	500	0.861±0.156	0.996	0.924	250	2500	0.943±0.066	0.996	0.980
25	600	0.855±0.152	0.996	0.908	250	3000	0.943±0.063	0.999	0.980
30	400	0.856±0.154	0.976	0.913	250	3500	0.939±0.066	1.000	0.964
30	500	0.856±0.157	0.996	0.909	275	2500	0.943±0.066	0.996	0.980
30	600	0.854±0.154	0.996	0.901	275	3000	0.934±0.061	1.000	0.978

Bolded values are optimal cut points for bedtime rest (CP₁) and activity (CP₂).

^a- Area under the ROC curve calculated as sensitivity multiplied by specificity before data were rounded;

^b- defined as the probability of correctly classifying bedtime rest period;

^c- defined as the probability of correctly classifying activity period.