Journal of Translational Medicine

Table 2 Predictive performance of the 9 selected SNPs in each of the 3 unseen Test datasets

From: Robust SNP-based prediction of rheumatoid arthritis through machine-learning-optimized polygenic risk score

Dataset	Evaluation metric	Machine learning models
Dataset	Evaluation metric	Logistic regression	Naïve bayes	Random forest	XGBoost	SVM RBF
Test 1 evaluation	AUC	0.990	0.986	0.993	0.994	0.991
	Sensitivity	0.928	0.936	0.928	0.936	0.936
	Specificity	0.963	0.960	0.963	0.963	0.963
	F1 score	0.913	0.914	0.913	0.918	0.918
	Accuracy	0.945	0.948	0.945	0.949	0.949
	Avg. Precision	0.970	0.951	0.975	0.976	0.970
Test 2 evaluation	AUC	0.988	0.982	0.989	0.992	0.990
	Sensitivity	0.937	0.945	0.945	0.945	0.937
	Specificity	0.961	0.961	0.963	0.963	0.963
	F1 Score	0.915	0.920	0.923	0.923	0.919
	Accuracy	0.949	0.953	0.954	0.954	0.950
	Avg. Precision	0.964	0.960	0.968	0.972	0.965
Test 3 evaluation	AUC	0.987	0.972	0.987	0.991	0.986
	Sensitivity	0.913	0.921	0.921	0.937	0.921
	Specificity	0.966	0.958	0.966	0.966	0.966
	F1 Score	0.910	0.903	0.914	0.922	0.914
	Accuracy	0.940	0.940	0.944	0.952	0.944
	Avg. Precision	0.961	0.936	0.958	0.965	0.946

Back to article page

ISSN: 1479-5876

Contact us

Submission enquiries: Access here and click Contact Us
General enquiries: info@biomedcentral.com