Abstract
Vitamin D is an essential nutrient closely associated with the prevention of multiple diseases, including osteoporosis, diabetes, and cardiovascular disorders. The serum level of 25-hydroxyvitamin D [25(OH)D] is the primary biomarker for assessing vitamin D status, and its precise quantification is critical for clinical diagnosis and treatment. Chemiluminescence immunoassay (CLIA) and liquid chromatography–tandem mass spectrometry (LC-MS/MS) are widely used analytical methods; however, methodological discrepancies often lead to inconsistent results that may be further influenced by demographic factors such as age and gender. In this study, we evaluated the consistency and correlation between CLIA and LC-MS/MS measurements using a Generative Fuzzy Inference System (GENFIS). Analysis of 138 serum samples showed that LC-MS/MS produced significantly higher 25(OH)D concentrations than CLIA (p < 0.01; mean difference = 1.33 ± 3.71; 95% CI: − 5.95 to 8.61), though the two methods exhibited strong linear correlation and agreement (Cohen’s Kappa = 0.8257; R² = 0.9075; intraclass correlation coefficient = 0.93). GENFIS analysis indicated a possible 30–40-year female pattern of larger between the detection differences of two methods. To verify this finding, an additional 59 samples were analyzed, revealing a relative risk (RR) of 3.18 (95% CI: 1.71–5.93; p < 0.05) for this subgroup compared with other populations, supporting the GENFIS inference. These results highlight age- and gender-related differences in 25(OH)D measurement between CLIA and LC-MS/MS, providing valuable insights for improving the standardization and clinical interpretation of vitamin D testing. SUPPLEMENTARY INFORMATION: The online version contains supplementary material available at 10.1038/s41598-026-41793-9.