Table 3. Comparison of the N-SRS, the R-FSRS, and other machine learning methods performance on the testing set of the Framingham datasets.

Reported metrics include sensitivity, specificity, precision, negative predictive value (NPV), and positive predictive value (PPV) at the probability threshold of 0.5. The Table also presents the overall c-statistic (AUC) and calibration χ² results.

	A) Framingham Dataset 1 (FD1)
	N-SRS	R-FSRS (both genders)	R-FSRS (men)	R-FSRS (women)	Log. Reg	CART	Random Forest	XGBoost
Sensitivity	0.9142	0.8510	0.8461	0.8554	0.8933	0.8802	0.9175	0.9167
Specificity	0.7238	0.6902	0.6890	0.7043	0.7102	0.7099	0.7161	0.7354
Precision	0.9408	0.9620	0.9353	0.9758	0.9701	0.9736	0.9605	0.9412
NPV	0.0592	0.0380	0.0647	0.0242	0.0423	0.0380	0.0863	0.0588
PPV	0.9408	0.9620	0.9353	0.9758	0.9621	0.9736	0.9137	0.9412
AUC	0.8743	0.7374	0.7188	0.7552	0.8065	0.7981	0.8829	0.8846
AUC 95% CI	0.8569–0.9014	0.6976–0.7619	0.6765–0.7636	0.7081–0.8102	0.772–0.8351	0.7676–0.8287	0.8578–0.9081	0.8643–0.9048
calibration χ2	1.96	8.05	11.98	5.44	2.88	3.04	1.43	1.58
	B) Framingham Dataset 2 (FD2)
	N-SRS	R-FSRS (both genders)	R-FSRS (men)	R-FSRS (women)	Log. Reg	CART	Random Forest	XGBoost
Sensitivity	0.8948	0.8533	0.8605	0.8487	0.8763	0.8504	0.8938	0.8934
Specificity	0.5097	0.4217	0.4066	0.4800	0.4867	0.2505	0.4994	0.5110
Precision	0.9693	0.9617	0.9531	0.9712	0.9688	0.9393	0.9816	0.9700
NPV	0.3973	0.2233	0.2321	0.1933	0.2576	0.1704	0.3804	0.4053
PPV	0.9693	0.9617	0.9531	0.9712	0.9401	0.9486	0.9535	0.9700
AUC	0.8238	0.7488	0.7281	0.7677	0.7754	0.6884	0.8216	0.8260
AUC (95% CI)	0.791–0.8558	0.7145–0.7831	0.6775–0.7788	0.7149–0.8204	0.738–0.8119	0.6435–0.7333	0.7881–0.8536	0.7938–0.8567
calibration χ2	2.75	7.3	12.1	4.1	6.5	20.34	2.81	2.7