. 2022 Feb 17;12:2738. doi: 10.1038/s41598-022-06585-x

Table 4.

Comparative chart displaying the accuracy score, area under the receiver operating characteristic curve, F1 score, and Matthews Correlation Coefficient (MCC) for each individual machine learning model.

Model	Accuracy	ROC AUC	F1	MCC
Logistic regression	88% (85–90%) p < 0.05	0.89 (0.86–0.92) p < 0.05	0.74 (0.69–0.78) p < 0.05	0.50 (0.42–0.58) p < 0.05
Random forest	88% (86–90%) p < 0.05	0.88 (0.84–0.91) p < 0.05	0.72 (0.67–0.77) p < 0.05	0.52 (0.44–0.60) p < 0.05
XGBoost	87% (85–90%) p < 0.05	0.88 (0.85–0.91) p < 0.05	0.73 (0.69–0.78) p < 0.05	0.51 (0.42–0.59) p < 0.05
K-nearest neighbors	81% (80–86%) p < 0.05	0.72 (0.67–0.76) p < 0.05	0.53 (0.49–0.58) p < 0.05	0.19 (0.09–0.28) p < 0.05
Support vector machine	81% (80–85%) p < 0.05	0.50 (0.44–0.56) p < 0.05	0.45 (0.44–0.46) p < 0.05	0.0 (0.00–0.00) p < 0.05

Model

Accuracy

ROC AUC

MCC

Logistic regression

88% (85–90%)

p < 0.05

0.89 (0.86–0.92)

p < 0.05

0.74 (0.69–0.78)

p < 0.05

0.50 (0.42–0.58)

p < 0.05

Random forest

88% (86–90%)

p < 0.05

0.88 (0.84–0.91)

p < 0.05

0.72 (0.67–0.77)

p < 0.05

0.52 (0.44–0.60)

p < 0.05

XGBoost

87% (85–90%)

p < 0.05

0.88 (0.85–0.91)

p < 0.05

0.73 (0.69–0.78)

p < 0.05

0.51 (0.42–0.59)

p < 0.05

K-nearest neighbors

81% (80–86%)

p < 0.05

0.72 (0.67–0.76)

p < 0.05

0.53 (0.49–0.58)

p < 0.05

0.19 (0.09–0.28)

p < 0.05

Support vector machine

81% (80–85%)

p < 0.05

0.50 (0.44–0.56)

p < 0.05

0.45 (0.44–0.46)

p < 0.05

0.0 (0.00–0.00)

p < 0.05

A 95% confidence interval is listed for each metric in parentheses, and a p-value expressing if a metric for each model is significantly different from the same metric from all the other models is listed.