Association between the albumin-bilirubin score and gallstone prevalence: evidence from NHANES and an independent hospital-based cohort.

Objective: This study aimed to evaluate the association between the albumin-bilirubin (ALBI) score and gallstone prevalence, and to develop machine-learning-based classification models for identifying participants with gallstones. Methods: Data from NHANES 2017-March 2020, including adults with complete albumin, bilirubin, and gallstone-status information, were analyzed. Weighted multivariable logistic regression and restricted cubic spline models were used to examine the association between ALBI score and gallstone prevalence. Fifteen feature-selection and machine-learning combinations were implemented to build classification models, with external validation conducted in a hospital-based cohort. Model performance was assessed using ROC curves, calibration plots, and decision curve analysis. Results: A total of 5,090 participants were included, including 490 with gallstones (9.63%, crude proportion). In the fully adjusted survey-weighted logistic regression model, continuous ALBI score was positively associated with gallstone prevalence (OR = 2.89; 95% CI: 1.68-4.99; Conclusion: In a national sample of US adults, a higher ALBI score remained associated with higher odds of gallstone disease after multivariable adjustment. Machine-learning models incorporating ALBI achieved moderate discrimination (cross-validated AUC up to 0.720; internal test AUC up to 0.708), with preliminary external-validation support that requires confirmation in larger cohorts.