Lasso Regression-based Model for Cross-Sectional Identification of Significant Hepatic Fibrosis in NAFLD LASSO-based identification of hepatic fibrosis in NAFLD.

Abstract

Objective: Nonalcoholic fatty liver disease (NAFLD) is an increasing global health concern, with liver-related mortality increasing as fibrosis progresses. This study aimed to identify the key determinants and develop a noninvasive model to detect significant hepatic fibrosis.

Methods: A total of 466 patients with biopsy-confirmed NAFLD were retrospectively analyzed at Beijing Ditan Hospital between 2008 and 2018. The patients were classified into non-significant (S0-1) and significant fibrosis (S2-4) groups. Relevant features were selected using least absolute shrinkage and selection operator (LASSO) regression, followed by multivariate logistic regression to construct a model for the cross-sectional identification of significant fibrosis. Model performance was assessed using receiver operating characteristic (ROC) curves, decision curve analysis (DCA), and bootstrap validation.

Results: Of the 466 patients, 112 had significant fibrosis. LASSO regression identified 10 relevant features, and the model achieved an AUC of 0.919 (sensitivity, 83.9%; specificity, 85.3%) with a corrected AUC of 0.907 after bootstrap validation. It outperformed the APRI, FIB-4, and LSM ( P < 0.001), and the DCA confirmed its clinical utility across probability thresholds.

Conclusion: The noninvasive model, incorporating demographic, laboratory, and imaging parameters, accurately identified significant hepatic fibrosis in NAFLD and outperformed existing noninvasive scores. This may facilitate interventions and guide personalized management.