Table 3.

Classification 3 performance (C-3) (Patients with NGT vs. Patients with prediabetes vs. Patients with T2D).

ML algorithms	Accuracy	Cohen Kappa	Mean Accuracy (SD, CV=10)	Mean Cohen Kappa (SD, CV=10)
Logistic Regression	0.77	0.61	0.76 (0.06)	0.65 (0.11)
Naive Bayes	0.71	0.5	0.68 (0.07)	0.49 (0.12)
Decision Tree	0.6	0.29	0.65 (0.08)	0.35 (0.17)
Random Forest	0.98	0.63	0.95 (0.03)	0.9 (0.03)
XGBoost	0.92	0.87	0.96 (0.02)	0.93 (0.05)
Multilayer Perceptron (MLP)	0.93	0.87	0.9 (0.06)	0.88 (0.07)

We compared the performance of six ML algorithms using the Precision and Cohen Kappa score values. We use the stratified cross-validation (CV) technique to obtain our results’ median and standard deviation (SD) (K Fold = 10).