Skip to main content

View full-text article in PMC

. 2023 Nov 24;11:e47859. doi: 10.2196/47859

Table 2.

Validation results obtained using the DT^a classifier: mean (SD) values of 5 experiments.

Data type, GAN^b, and approach		NSCLC^c		Breast cancer		Diabetes
Data type, GAN^b, and approach		AUC^d	F₁-score	AUC	F₁-score	AUC	F₁-score
OD^e		66.06 (1.31)	66.11 (1.30)	61.14 (1.66)	49.64 (1.06)	65.58 (0.13)	47.82 (0.04)
Balanced STD^f
	CTGAN^g, no condition	63.46 (1.91)	61.87 (1.83)	59.10 (2.17)	62.20 (1.93)	57.19 (0.25)	49.16 (0.10)
	CTGAN, CS^h	59.95 (1.26)	58.65 (2.44)	64.65 (1.66)	56.67 (1.31)	58.93 (0.18)	47.89 (0.09)
	CTGAN, DCⁱ	74.50 (1.23)	71.99 (0.77)^j	73.31 (1.11)^j	67.10 (1.83)	61.57 (0.09)^j	50.45 (0.05)
	CopulaGAN^k, no condition	66.11 (1.32)	65.16 (1.27)	59.82 (1.69)	52.32 (2.11)	58.93 (0.24	47.13 (0.04)
	CopulaGAN, CS	63.87 (2.02)	63.07 (1.87)	67.96 (2.15)	62.20 (1.93)	60.08 (0.17)	49.77 (0.04)
	CopulaGAN, DC	74.87 (0.77)^j	70.54 (0.76)	70.66 (0.85)	68.92 (1.78)^j	60.61 (0.06)	53.80 (0.08)^j
Imbalanced STD
	CTGAN, no condition	50.93 (0.89)	44.99 (1.97)	52.48 (0.09	38.75 (0.51	57.60 (0.14	47.09 (0.01
	CTGAN, CS	50.64 (0.80)	44.80 (1.48)	52.48 (0.09)	38.75 (0.51)	58.34 (0.24)	47.09 (0.01)
	CTGAN, DC	57.99 (2.06)	57.78 (2.81)	56.00 (0.31)	38.75 (0.51)	60.52 (0.05)	50.36 (0.10)
	CopulaGAN, no condition	52.32 (1.05)	48.38 (1.92)	53.81 (0.63)	38.75 (0.51)	57.10 (0.21)	47.09 (0.01)
	CopulaGAN, CS	52.06 (0.91)	47.62 (1.70)	55.50 (1.01)	38.75 (0.51)	56.21 (0.24)	47.09 (0.01)
	CopulaGAN, DC	55.86 (3.10)	54.12 (4.60)	57.95 (0.76)	38.75 (0.51)	59.38 (0.15)	50.22 (0.08)

^aDT: decision tree.

^bGAN: generative adversarial network.

^cNSCLC: non–small cell lung cancer.

^dAUC: area under the curve.

^eOD: original data.

^fSTD: synthetic tabular data.

^gCTGAN: conditional tabular generative adversarial network.

^hCS: conditional sampling.

ⁱDC: divide and conquer.

^jThe best results.

^kCopulaGAN: copula generative adversarial network.