Table 6. Comparison of classifier performance using FC and dFC.

Cross validation
Study		Dataset	Controls	Cases	# features	AUC	B Accuracy	Sensitvity	Specificity	Comments
Lei et al. (2020)	FC	COBRE	72	68	4,095		0.82	0.69	0.94
Morgan et al. (2021)	FC	COBRE	73	60	42,778	0.75	0.70	0.62	0.77
Hancock et al. (2022)	dFC	COBRE	71	59	1	0.71	0.62	0.64	0.60	Case Positive
										Downsampled
Morgan et al. (2021)	FC	Maastricht	53	59	42,778	0.74	0.65	0.77	0.59
		Dublin	72	25	42,778	0.82	0.86	0.50	0.97
Rashid et al. (2016)	dFC	Hartford	135	87	15		0.84	0.83	0.92
Du et al. (2020)	dFC	BSNIP-1	238	113	>1’000		0.69	0.66	0.73
Hancock et al. (2022)	dFC	HCPEP	53	82	1	0.73	0.67	0.59	0.73	Case Positive
										Downsampled
External validation
Study		Dataset	Controls	Cases	# features	AUC	B Accuracy	Sensitvity	Specificity
Morgan et al. (2021)	FC	Maastricht->Dublin	53	59	42,778	0.77	0.56
		Dublin->Maastricht	72	25	42,778	0.76	0.69
Hancock et al. (2022)	dFC	COBRE->HCPEP	53	53	1	0.76	0.57	0.93	0.21	Case Positive
		HCPEP->COBRE	53	53	1	0.71	0.58	0.19	0.96	Case Positive
										Random Sampling