Associations between alkaline phosphatase levels and abdominal aortic calcification via statistical logistic modeling and machine learning

Background: Vascular calcification is associated with serum alkaline phosphatase (ALP) levels. This study aimed to investigate the relationship between ALP and abdominal aortic calcification (AAC). Methods: Data from the 2013–2014 National Health and Nutrition Examination Survey (NHANES) were utilized for this study. Weighted multivariable regression analyses and subgroup analyses were conducted to investigate the independent relationship between ALP levels and AAC scores. Restricted cubic splines (RCS) regression was employed to explore nonlinear relationships. Additionally, Boruta, random forest, Lasso, and logistic combined with SHAP interpretable analysis were utilized for screening variables and ranking their importance. Results: A total of 1,605 middle-aged adult participants were included in the analysis. ALP was positively associated with AAC scores among middle-aged females (β=0.008, 95% CI: 0.002–0.014, P=0.011). The risk of AAC was 3.04 times greater (OR=3.045, 95% CI: 1.019--9.098, P=0.047) in middle-aged females (≥93 U/L vs. <59 U/L). A significant relationship between ALP levels and AAC was observed among middle-aged females, with one notable inflection point identified at an ALP level of 63 U/L in the RCS regression. All four machine learning algorithms interpret ALP as one of the most significant features infecting AAC among middle-aged females. Conclusions: High ALP levels were associated with elevated AAC scores and an increased risk of AAC among middle-aged females. Abdominal CT or abdominal X-ray examinations are recommended when the ALP level exceeds 63 U/L during routine biochemical assessments.