Table 2.

Performance metrics of machine learning models across 5-fold cross-validation data for periodontitis severity.

	XGBoost	Random forest	Logistic regression	Support vector machine	Multilayer perceptrons	Reduced model XGBoost
Precision	0.6529 ± 0.0379	0.6319 ± 0.0306	0.6179 ± 0.0258	0.5115 ± 0.0246	0.5025 ± 0.0258	0.5928 ± 0.0232
Recall	0.6145 ± 0.0488	0.5778 ± 0.0351	0.5203 ± 0.0267	0.4335 ± 0.0204	0.3346 ± 0.0240	0.5757 ± 0.0247
Area under curve	0.8151 ± 0.0270	0.7974 ± 0.0301	0.7557 ± 0.0314	0.4241 ± 0.0238	0.2319 ± 0.0084	0.7613 ± 0.0282
F1	0.6124 ± 0.0449	0.5791 ± 0.0366	0.5282 ± 0.0274	0.5707 ± 0.0471	0.5816 ± 0.0314	0.5572 ± 0.0228

The reduced model was defined as the model excluding thyroid function parameters as predictors.

This model incorporated only the following covariates: sex, age, BMI, smoking status, hypertension status, educational attainment, overweight status, and vitamin D levels.