Diagnostic efficacy of remnant cholesterol inflammatory index in diabetic kidney disease: machine learning approaches

Background Emerging evidence indicates that remnant cholesterol (RC) and inflammation play a crucial role in diabetic kidney disease (DKD) pathogenesis. The association and diagnostic efficacy of remnant cholesterol inflammatory index (RCII), integrating RC and inflammatory markers, with DKD remains underexplored. Methods This cross-sectional study analyzed data from the National Health and Nutrition Examination Survey (NHANES) 2015–2020, including 5,943 participants. DKD was defined by diabetes, urine albumin to creatinine ratio (ACR) ≥ 30 mg/g and an estimated glomerular filtration rate (eGFR) < 60 mL/min/1.73 m 2 . RC was calculated as total cholesterol minus high-density and low-density lipoprotein cholesterol, while RCII was derived by multiplying RC by high-sensitivity C-reactive protein (hs-CRP). Logistic regression and restricted cubic spline analysis were used to evaluate associations and dose–response relationship between RC and RCII and DKD. We assessed RCII diagnostic efficacy measured by five machine learning algorithms. Results Our study observed 1,014 cases of DKD (17.06%), with a higher prevalence among males (14.1%) compared to females (11.7%). The highest RC (OR: 2.73, 95% CI: 2.12–3.52, P for trend<0.001) and RCII (OR: 2.29, 95% CI: 1.77–2.97, P for trend <0.001) levels were significantly associated with increased DKD risk after full adjustment. The result showed both overall and nonlinear positive correlations between the risk of DKD and both RC ( P for overall <0.001, P for nonlinear = 0.049) and RCII ( P for overall <0.001, P for nonlinear <0.001). Machine learning models incorporating RCII and traditional risk factors demonstrated robust diagnostic efficacy, with extreme gradient boosting (XGBoost) achieving the highest AUC values in the testing set (AUC: 0.953). Conclusion Our study suggested RCII was a novel and promising biomarker for DKD risk. Its integration into diagnostic models may enhance early identification and personalized prevention strategies for DKD, addressing a critical need in diabetes management.