Table 3.

Comparison of the diagnostic performance of TB assessment using different data inputs and algorithms based on biomarkers.

	Machine Learning	Cohorts	AUC [95% CI]	Sensitivity	Specificity
Biomarker	LR	Validation	0.978 (0.964‐0.986)	0.850	1.000
		Discovery	0.970 (0.948‐0.984)	0.967	0.938
	PLS	Validation	0.968 (0.952‐0.980)	0.900	0.946
		Discovery	0.986 (0.978‐0.993)	0.967	0.963
	RF	Validation	0.953 (0.926‐0.968)	0.850	0.946
		Discovery	1.000 (1.000‐1.000)	1.000	1.000
	SVM	Validation	0.960 (0.938‐0.976)	0.900	0.912
		Discovery	0.978 (0.963‐0.990)	0.956	0.963