Table 4.

Test set performance of each model with various data set combinations.

Feature	Model	Accuracy	AUROCa	Specificity	Sensitivity	Precision	F1 score
All features	Random forest	0.813	0.872	0.938	0.574	0.827	0.677
Lifestyle and environment	RGFb	0.674	0.687	0.773	0.477	0.513	0.495
Questionnaire	RGF	0.771	0.843	0.858	0.617	0.712	0.661

^aAUROC: area under the receiver operating characteristic.

^bRGF: regularized greedy forest.