Skip to main content

View full-text article in PMC

. 2020 Jun 8;8(6):e17364. doi: 10.2196/17364

Table 3.

Performance of the four machine learning algorithms on predicting breast cancer risk.

Algorithm	AUC^a, mean (SD)	Sensitivity^b, mean (SD)	Specificity^b, mean (SD)	Accuracy^b, mean (SD)
Extreme gradient boosting	0.742 (0.009)	0.656 (0.017)	0.686 (0.012)	0.671 (0.009)
Random forest	0.728 (0.009)	0.650 (0.016)	0.677 (0.015)	0.663 (0.010)
Deep neural network	0.728 (0.010)	0.642 (0.037)	0.679 (0.033)	0.661 (0.010)
Logistic regression	0.631 (0.008)	0.496 (0.020)	0.661 (0.021)	0.578 (0.008)

^aAUC: area under the receiver operating characteristic curve.

^bSensitivity, specificity, and accuracy were calculated using the default cutoff value (0.5).