Fig. 4

Feature selection using the LASSO regression method. a The 5- fold cross validation was used to select the optimal parameters (Lambda, λ = 0.020). b The AUC was plotted versus log (Lambda) by using the minimum standard (left line) and the 1 standard error of minimum standard (1 − SE standard, right line) to draw the vertical line with the best value. The minimum standard was used according to the 5-fold cross validation. Eleven features were chosen after LASSO regression