Table 9. Accuracy, cohen’s κ, TPR and TNR values based for the all ML tools and using either 5-fold cross validation or 20% testing set for prediction people with T2DM from people without T2DM.

ML tool	Type of Data	Measure
		Accuracy	Cohen’s	TPR	TNR
Logistic	5-fold cross validation	0.75	0.50	0.76	0.74
	20% Testing	0.97	0.94	1.00	0.95
RF	5-fold cross validation	0.76	0.51	0.79	0.73
	20% Testing	0.66	0.34	0.97	0.38
XGBoost	5-fold cross validation	0.73	0.64	0.72	0.73
	20% Testing	0.66	0.33	0.94	0.41
PNN	5-fold cross validation	0.62	0.24	0.60	0.64
	20% Testing	0.86	0.71	0.88	0.84
C-LibSVM	5-fold cross validation	0.65	0.31	0.71	0.60
	20% Testing	0.47	0.00	1.00	0.00
nu-LibSVM	5-fold cross validation	0.66	0.32	0.68	0.64
	20% Testing	0.53	0.00	1.00	0.00
AdaBoost	5-fold cross validation	0.74	0.49	0.82	0.61
	20% Testing	0.60	0.23	0.97	0.27
Gradient-boost	5-fold cross validation	0.73	0.46	0.71	0.75
	20% Testing	0.59	0.18	0.64	0.54
KNN*	5-fold cross validation	0.62	0.24	0.54	0.70
	20% Testing	0.47	0.00	1.00	0.00
K-star	5-fold cross validation	0.73	0.46	0.68	0.78
	20% Testing	0.53	0.07	0.67	0.41

* Features with non-numeric values are ignored.