R 2 was used to backward-select each combination of covariates with statistical significance (p-values <0.1): the covariates that did not reach statistical significance were removed until R 2 remained the same, improved, or decreased the least when all possible combinations of remaining covariates were considered.
1
—
—