A Fact-Finding Procedure Integrating Machine Learning and AHP Technique to Predict Delayed Diagnosis of Bladder Patients with Hematuria

Table 4

Performance results of classifiers.

Classifier	Algorithms	Accuracy, μ/σ	Sensitivity, μ/σ	Specificity, μ/σ	AUC, μ/σ
Without AdaBoost	C4.5	0.859/0.014	0.843/0.003	0.858/0.016	0.871/0.042
	RF	0.879/0.071	0.875/0.045	0.8720.081	0.942/0.048
	SVM	0.746/0.007	0.752/0.005	0.769/0.040	0.705/0.008
	LGR	0.788/0.011	0.799/0.013	0.802/0.021	0.854/0.011
	MLP	0.742/0.079	0.720/0.048	0.709/0.127	0.775/0.104
With AdaBoost	C4.5	0.856/0.088	0.825/0.109	0.856/0.088	0.915/0.064
	RF	0.881/0.066	0.857/0.062	0.881/0.066	0.943/0.045
	SVM	0.743/0.011	0.722/0.008	0.743/0.001	0.762/0.010
	LGR	0.791/0.002	0.802/0.025	0.791/0.003	0.828/0.008
	MLP	0.751/0.088	0.791/0.029	0.752/0.087	0.786/0.111