Figure 2. Variable selection with the least absolute shrinkage and selection operator (LASSO) regression model. (A) 10-fold cross validation in the LASSO regression model. The solid vertical line with a red dot represents the cross-validation curve and the SE of partial likelihood deviance. The vertical dot line represents the optimal lambda value. The lambda of 0.00293 was selected for the LASSO regression. (B) The coefficient profiles of the 21 candidate variables. The L1 norm is a regularisation term to prevent overfitting problems. Each coloured line represents a candidate variable.