Use of Machine Learning and Routine Laboratory Tests for Diabetes Mellitus Screening

Table 3

Performance of the classification models using healthy and no healthy (HN) and no diabetes and diabetes (ND) datasets. Percentual (%) values of sensitivity (SN), specificity (SP), precision (PR), negative precision (NPR), and F1 score (F1).

	Healthy and no healthy (%)					No diabetes and diabetes (%)
	KNN	SVM	NB	RF	ANN	KNN	SVM	NB	RF	ANN
SN	55.9	77.3	57.6	78.2	78.1	42.7	59.9	69.6	66.3	67.9
SP	77.5	77.5	88.1	75.3	76.6	99.5	98.7	95.1	98.1	97.9
PR	76.6	79.0	84.2	77.7	78.7	93.7	89.0	71.0	85.7	84.8
NPV	68.1	75.6	65.4	75.8	76.1	91.0	93.4	94.8	94.4	94.7
F1	71.4	78.2	68.4	77.9	78.4	58.7	71.3	70.3	74.7	75.4