Research Article

Explainable Machine Learning-Based Prediction Model for Diabetic Nephropathy

Table 2

Highest AUC scores achieved by hyperparameter tuning of four machine learning models.

ClassifiersAdjusted hyperparametersAccuracyRecallAUC value

XGB, , , , , , , , 0.8750.8750.966≤0.001
RF, , min_samples_leaf = 2, 0.8750.8740.937≤0.001
DT, , 0.8930.8930.812≤0.001
Logistic, , 0.7550.750.845≤0.001

C indicates the inverse of the regularization factor; tol indicates that the calculation stops when the solution reaches 0.0001 and the optimal solution is considered to have been found; values are the Delong’s test for the area under the receiver operating characteristic curve, and <0.05 indicates that the analysis method is statistically significant. Abbreviations: AC: abdominal circumference; ALP: alkaline phosphatase; AGI: alpha-glucosidase inhibitor; Apo AI: apolipoprotein A1; Apo B: apolipoprotein B; AUC: area under the curve; BMI: body mass index; C2: acetylcarnitine; C4DC: succinylcarnitine; C5DC: glutarylcarnitine; C24: tetracosanoic carnitine; CHOL: cholesterol; Cys: cysteine; DN: diabetic nephropathy; DT: decision tree; DBP: diastolic blood pressure; DCA: decision curve; Duration of T2D: duration of type 2 diabetes mellitus; drink: alcohol consumption or not; Dpp-4: dipeptidyl peptidase-4; eGFR: glomerular filtration rate; FT3: free triiodothyronine; FBG: fasting blood glucose; FBG: fasting blood glucose; GLB: globulin; GADA: glutamic acid decarboxylase antibody; GLP-1: glucagon-like peptide; HDL-C: high-density lipoprotein cholesterol; HbA1c: glycated hemoglobin; HB: hemoglobin concentration; HbA1c: glycosylated hemoglobin; IGF-1: insulin sample growth factor-1; INS: insulin; LASSO: least absolute shrinkage and selection operator; Met: methionine; MAU: urinary microalbumin; NNs: artificial neural networks; PCV: erythrocyte pressure volume; ROC:, receiver operating characteristic; RF: random forest; SBP: systolic blood pressure; SCR: serum creatinine; SGLT-2: sodium-glucose cotransporter protein 2 inhibitor; Ser: serine; SVM: support vector machine; Tyr: chitosan; TSH: thyroid-stimulating hormone; TZDs: thiazolidinediones; UA: blood uric acid; XGB: extreme gradient boosting.