Supervised Machine Learning Based Noninvasive Prediction of Atrial Flutter Mechanism from P-to-P Interval Variability under Imbalanced Dataset Conditions

Table 6

Mean value of classifier performance at 400% data augmentation.

Technique	Performance	LDA	LOG	SVM
SMOTE	Accuracy	72.48	75.22	72.76
	Specificity	28.38	41.42	23.30
	Sensitivity	94.29	91.54	96.81
Modified-SMOTE	Accuracy	73.86	75.82	72.89
	Specificity	29.51	39.70	21.41
	Sensitivity	95.20	93.76	97.85
Smoothed-bootstrap	Accuracy	72.96	76.13	71.80
	Specificity	27.94	40.36	15.93
	Sensitivity	94.83	93.56	99.13