Evaluation of an accelerometer-based monitor for detecting bed net use and human entry/exit using a machine learning algorithm

Table 4 Performance of three-category classification model for bed net use behaviours in validation dataset

Net use behaviour	Adults				Children				Comparison of AUC
Net use behaviour	Observations	Sensitivity	Specificity	AUC (95% CI)	Observations	Sensitivity	Specificity	AUC (95% CI)	Comparison of AUC
Sleep	1115	0.995	0.967	0.994 (0.985–1.000)	99	1.000	0.951	0.999 (0.995–1.000)	p = 0.323
Enter net	362	0.636	0.946	0.934 (0.907–0.962)	99	0.556	0.810	0.798 (0.684–0.911)	p = 0.026*
Exit net	363	0.730	0.902	0.927 (0.900–0.953)	99	0.565	0.784	0.800 (0.689–0.909)	p = 0.030*
		Overall accuracy		318/368 (86.4%)		Overall accuracy		42/60 (70.0%)

ISSN: 1475-2875