Predicting colorectal adenomatous polyps in patients with chronic liver disease: A novel nomogram

Figure 2 Least absolute shrinkage and selection operator regression analysis identified potential predictive factors. A: The least absolute shrinkage and selection operator model was optimized by selecting parameters based on the criteria method, which added one standard error to the minimum error. This process utilized five-fold cross-validation, resulting in λ = 0.03902279 and log(λ) = -3.241, satisfying the one standard error criteria for a simpler and more regularized model; B: The regression coefficients varied as λ increased, smaller λ values were associated with larger coefficients for most features, while larger λ values resulted in coefficients being reduced towards zero or reaching zero, a total of nine variables with non-zero coefficients were identified.