. 2024 Mar 17;14:6404. doi: 10.1038/s41598-024-57234-4

Table 2.

Accuracies of genomic prediction and root mean square error (RMSE) assessed for feed efficiency-related traits in testing animals using a forward validation scheme obtained with different parametric and machine learning methods.

Trait¹	Model fit	Genomic prediction approaches²
Trait¹	Model fit	STGBLUP	BayesA	BayesB	BayesC	BL	BRR	MTGBLUP	MLNN	SVR
ADG (kg/day)	r—training	0.88	0.96	0.89	0.93	0.95	0.91	0.87	0.88	0.86
	Accuracy	0.58	0.53*	0.59	0.53*	0.54*	0.59	0.66***	0.62**	0.67***
	RMSE	0.061	0.063	0.061	0.062	0.071	0.061	0.048	0.054	0.047
DMI (kg/day)	r—training	0.86	0.94	0.92	0.92	0.9	0.87	0.83	0.85	0.83
	Accuracy	0.56	0.54*	0.57	0.55	0.56	0.57	0.62***	0.60***	0.62***
	RMSE	0.365	0.371	0.366	0.373	0.444	0.365	0.276	0.327	0.271
FE	r—training	0.95	0.97	0.94	0.9	0.89	0.88	0.84	0.86	0.83
	Accuracy	0.53	0.54	0.56	0.54	0.57*	0.56*	0.64***	0.61***	0.64***
	RMSE	0.007	0.007	0.007	0.006	0.008	0.006	0.0051	0.006	0.0049
RFI (kg/day)	r—training	0.89	0.91	0.93	0.9	0.92	0.89	0.86	0.89	0.86
	Accuracy	0.62	0.63	0.64	0.62	0.63	0.63	0.68***	0.66*	0.69***
	RMSE	0.231	0.235	0.23	0.235	0.284	0.233	0.174	0.209	0.167

¹ADG: average daily gain, DMI: dry matter intake, FE: feed efficiency, and RFI: residual feed intake, r: training— $c o r (y^{*}, {\hat{y}}_{i}^{*})$ in training population, Accuracy—c $o r (y^{*}, {\hat{y}}_{i}^{*})$ .

²STGBLUP: single trait GBLUP, BayesA: Bayesian A, BayesB: Bayesian B, BayesC: Bayesian C, BL: Bayesian Lasso, BRR: Bayesian ridge regression, MTGBLUP: multi-trait GBLUP, MLNN: Multi-layer neural networks, and SVR: support vector machine regression using a radial basis kernel. Statistically significant differences between the predictive ability of each method across the herds compared to the standard model (STGBLUP) were *p-value < 0.05, ** p-value < 0.01 and ***p-value < 0.005.