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

Table 2

Performance of the studied classification models using healthy, prediabetes, and diabetes (HPD) dataset.

Class	KNN	SVM	NB	RF	ANN
Sensitivity (SN %)
Diabetes	47.0	62.4	63.1	62.4	66.2
Prediabetes	53.4	63.7	56.3	63.7	67.9
Healthy	78.7	79.2	79.4	79.2	76.6
Specificity (SP %)
Diabetes	99.0	98.4	95.9	98.4	98.0
Prediabetes	74.2	76.5	77.8	76.5	75.0
Healthy	64.6	75.0	72.6	75.9	79.1
Precision (PR %)
Diabetes	89.1	86.9	72.3	86.9	84.9
Prediabetes	55.7	62.2	50.6	62.2	62.2
Healthy	66.9	74.2	72.5	74.2	76.9
Negative precision (NPR %)
Diabetes	91.6	93.8	93.8	93.8	94.4
Prediabetes	72.5	77.7	74.6	77.7	79.4
Healthy	76.9	79.8	79.5	79.8	78.8
F1 score (F1 %)
Diabetes	61.8	72.6	67.4	72.6	74.4
Prediabetes	54.5	62.6	58.3	62.9	64.9
Healthy	72.3	76.6	75.8	76.6	76.7