Fig. 2

The LASSO regression illustrates the effect of regularization on predictor coefficients via a coefficient path plot (a), identifies the optimal penalty strength through cross-validation (b), and ranks clinical variables by importance based on absolute regression coefficients (c). LASSO, Least Absolute Shrinkage and Selection Operator; DM, diabetes mellitus; HbA1c, glycated hemoglobin A1c; HDL-C, high-density lipoprotein cholesterol; BMI, body mass index.