Table 1. Model comparison and ranking using the LOO-CV method.

The ordering in which the models are displayed corresponds to the performance of cross-validation for CVA6.

	Time-varying FOI?	Age-dependent FOI?	Seroreversion?	EV-A71		CVA6
Model				elpd_diff	se_diff	elpd_diff	se_diff
5	No	Yes	No	-1.1	4.5	0	0
2	No	No	Yes	-4.0	4.5	-2.0	1.6
6	No	Yes	Yes	-1.5	6.3	-2.9	1.4
4	Yes	No	Yes	0	0	-3.4	2.6
3	Yes	No	No	-23.2	11.2	-44.0	12.1
1	No	No	No	-260.4	44.4	-260.9	47.8

Model comparison and ranking of the fitted catalytic models using the approximate leave-one-out cross-validation (LOO-CV) method. LOO-CV calculates the expected log pointwise predictive density (ELPD) for each model, which is a measure of the overall model fit accounting for model complexity. Model ranking is based on the differences in the ELPD and standard error estimates (‘elpd_diff’, and ‘se_diff’, respectively), where the differences are calculated relative to the model with the largest ELPD.