Prediction of the effects of small molecules on the gut microbiome using machine learning method integrating with optimal molecular features

Table 3 Results from the five-fold cross-validation and external validation of the 13MD-CM

	SE (%)	SP (%)	ACC (%)	MCC	AUC	F1-scorce
Five-fold cross-validation	67.4 ± 1.8	90.8 ± 0.8	82.9 ± 0.7	0.608 ± 0.017	0.791 ± 0.009	0.725 ± 0.014
External validation	60.7 ± 6.7	91.7 ± 3.3	82.2 ± 2.6	0.554 ± 0.058	0.759 ± 0.03	0.669 ± 0.05

ISSN: 1471-2105