Table 3.

Age prediction performances of the different statistical models on an independent validation set.

Model	Number of CpGs	CpGs	Estimators	Training set (n = 1,028)	Testing set 1 (n = 385)	Independent testing set 2 (n = 100)
						1 PCR and 1 PSQ/PCR (1 replicate)	1 PCR and 2 PSQ/PCR (2 replicates)	2 PCR and 1 PSQ/PCR (2 replicates)	3 PCR and 1 PSQ/PCR (3 replicates)	3 PCR and 2 PSQ/PCR (6 replicates)
Zbiec-Pierkarska 1	2	CpG5,7	R	0.918	0.932	0.880	0.893	0.902	0.909	0.914
			MAD	6.885	6.397	5.445	5.319	5.147	5.050	5.011
			RMSE	9.127	8.803	6.870	6.624	6.440	6.290	6.201
MQR	4	CpG6 & CpG42,62–72	R	0.934	0.945	0.904	0.911	0.919	0.924	0.927
			MAD	5.521	4.910	4.786	4.619	4.425	4.266	4.232
			RMSE	7.574	7.057	6.225	5.996	5.765	5.598	5.504
SVMr	2	CpG6,7	R	0.947	0.948	0.902	0.906	0.917	0.923	0.925
			MAD	5.051	4.701	4.784	4.668	4.388	4.211	4.174
			RMSE	6.843	6.833	6.287	6.140	5.771	5.581	5.515
SVMl	2	CpG6,7	R	0.905	0.927	0.902	0.905	0.917	0.922	0.923
			MAD	6.246	6.036	5.536	5.484	5.289	5.211	5.197
			RMSE	9.078	8.095	6.874	6.796	6.525	6.404	6.375
BGR	2	CpG6,7	R	0.976	0.947	0.900	0.904	0.913	0.919	0.920
			MAD	3.471	4.772	4.892	4.842	4.577	4.436	4.469
			RMSE	4.660	6.931	6.397	6.314	5.973	5.803	5.741
mMDA	1	CpG6	R	0.906	0.927	0.902	0.905	0.917	0.922	0.923
			MAD	6.291	6.079	5.926	5.875	5.736	5.673	5.598
			RMSE	9.008	8.104	7.234	7.158	6.932	6.826	6.772