Objectives To evaluate whether quantitative tongue diagnosis indicators can predict mild renal function decline and to compare their predictive ability using logistic regression and machine-learning models.Methods In 2,009 adults who underwent health screening, tongue-image indicators (coating, light-red index, blue-purple index, tooth-mark index), body composition, and metabolic markers were analyzed.estimated glomerular filtration rate (eGFR) was dichotomized (90 vs. <90).Logistic regression, random forest, and XGBoost models were used to evaluate predictive performance.Results The blue-purple index was independently associated with reduced eGFR (odds ratio=3.3).Age and body composition (fat mass, fat percentage) were also significant predictors.Machine-learning models showed modest discrimination (area under the receiver operating characteristic curve 0.64 to 0.70), with the logistic model performing best.Coating and light-red indices showed limited predictive power.Conclusions Tongue color-particularly the blue-purple index-may serve as a noninvasive indicator related to early renal function decline, although overall predictive power remains moderate. (
Lee et al. (Sat,) studied this question.