Development and Validation of an Online Dynamic Nomogram for Predicting Nonobese Metabolic Dysfunction-associated fatty liver disease Based on Body Composition Analysis

Background The factors influencing nonobese Metabolic dysfunction-associated fatty liver disease are discussed, and an online dynamic nomogram model is constructed. Methods A retrospective review was conducted on medical history data from 216 patients diagnosed with nonobese metabolic dysfunction-associated fatty liver disease (MAFLD) and 322 nonobese normal individuals at the Second Affiliated Hospital of Soochow University. An automated random grouping procedure employing statistical software allocated the 538 subjects into training and verification cohorts at a 7:3 ratio. Initial screening of relevant indicators employed univariate and correlation analyses. Subsequently, significant potential independent risk factors (P < 0.05) were identified through the Lasso regression method, followed by cross-validation. These statistically significant indicators were further analyzed using binary logistic regression. Their discriminative capacity was evaluated using receiver operating characteristic (ROC) curves and the corresponding area under the curve (AUC). Finally, a dynamic online diagnostic nomogram was constructed. The nomogram's predictive performance was assessed via AUC, while its calibration was evaluated using a calibration plot. Results Six independent risk factors, namely, past history (odds ratio [OR]: 2.399, P = 0.0008), TG (OR: 1.176, P = 0.008), HDL-C (OR: 0.173, P = 0.014), LDL-C (OR: 3.916, P = 0.001), fat (OR: 4.299, P = 0.0009) and dPhaseAngle(OR: 3.174, P = 0.022), were screened from the results of Lasso regression method and a binary logistic regression analysis of the training cohort and included in the nonobese MAFLD online diagnostic nomogram. The nomogram predicted nonobese MAFLD with AUC values of 0.863 in the training cohort, and 0.83 in the validation cohort. The calibration curve also revealed that the nomogram predicted outcomes were close to the ideal curve, and that the predicted outcomes were consistent with the real outcomes. Conclusion An online dynamic nomogram for nonobese MAFLD patients with good predictive performance was constructed, which can be used as a practical approach for personalized early screening and auxiliary diagnosis of potential risk factors and can assist physicians in making personalized diagnoses and treatments for patients.