Detection of diabetic patients in people with normal fasting glucose using machine learning

Table 2 Comparison of performance of four machine learning methods

	AUC	PR-AUC	Recall	Specificity	Balanced accuracy
RF	0.860	0.071	0.017	0.999	0.508
SVM	0.801	0.051	0.333	0.942	0.638
DNN	0.844	0.082	0.417	0.948	0.685
LR	0.899	0.106	0.850	0.811	0.830

The values of AUC, PR-AUC, recall, specificity, and balanced accuracy are range from 0 to 1, with a higher value indicating a better performance
RF random forest, SVM supported vector machine, DNN deep neural network, LR logistic regression, AUC area under the curve of receiver operating characteristic, PR-AUC are under the curve of precision-recall

ISSN: 1741-7015