Objective To investigate the application value of a prediction model constructed by magnetic resonance elastography (MRE) in identifying high-risk state of significant liver fibrosis in patients with chronic hepatitis B (CHB).

Methods The retrospective cohort study was constructed. The clinical data of 94 patients with CHB who were admitted to The Second Affiliated Hospital of Chongqing Medical University from May 2024 to July 2025 were collected. There were 73 males and 21 females, aged 48 (38,55) years. Of the 94 patients, 53 were classified as being in the high-risk state of significant hepatic fibrosis, and 41 in the low-risk state. Observation indicators: (1) analysis of factors influencing high-risk state of significant liver fibrosis in patients; (2) construction and evaluation of a prediction model in predicting high-risk state of significant liver fibrosis in patients. Comparison of measurement data with skewed distribution between groups was conducted by using the Mann-Whitney U test. Comparison of count data between groups was conducted by using the chi-square test. Univariate and multivariate analyses were conducted by using Logistic regression model. The model discrimination was evaluated through five-fold cross-validation, and receiver operating characteristic (ROC) curve was plotted to assess the diagnostic value of the model based on the area under the curve (AUC), sensitivity, and specificity. A calibration curve was plotted using 1 000 bootstrap self-sampling, and the fit was evaluated using the Hosmer-Lemeshow test. The consistency of the nomogram prediction model was evaluated using the calibration curve, and the clinical benefit was evaluated using the decision curve.

Results (1) Analysis of factors influencing high-risk state of significant liver fibrosis in patients: results of multivariate analysis showed that MRE stiffness value and liver apparent diffusion coefficient (ADC) value were independent influencing factors in identi-fying high-risk state of significant liver fibrosis in patients (odds ratio=6.737, 0.001, 95% confidence interval as 2.164‒20.978, 0.000‒0.284, P<0.05). (2) Construction and evaluation of a prediction model in predicting high-risk state of significant liver fibrosis in patients: results of ROC curve for MRE stiffness value showed that the optimal cutoff value was 3.955 kPa, the AUC was 0.791 (95% confidence interval as 0.687‒0.895), sensitivity was 0.962, specificity was 0.610, and the Yoden index was 0.572. Results of ROC curve for liver ADC value showed that the optimal cutoff value was 1.175×10^-3 mm²/s, the AUC was 0.753 (95% confidence interval as 0.654‒0.852), sensitivity was 0.736, specificity was 0.732, and the Yoden index was 0.468. Based on the results of multivariate analysis, a combined model was constructed by using the MRE stiffness value and liver ADC value in predicting high-risk state of significant liver fibrosis in patients, and results of ROC curve analysis showed that the AUC was 0.810 (95% confidence interval as 0.713‒0.907), the sensitivity was 0.868, the specificity was 0.732, and the Yoden index was 0.600. Results of calibration curve indicated a good fit between the predicted probabilities of the combined model and the actual occurrence probabilities. Results of Hosmer⁃Lemeshow test showed that χ²=10.049, df=8, P>0.05, suggesting good calibration of the model. Results of decision curve analysis showed that within the common threshold probability range, the combined model achieved high net benefit, indicating its high clinical application value.

Conclusion Both the MRE stiffness value and liver ADC value are independent influencing factors in identifying high-risk state of significant liver fibrosis in patients with CHB. These indicators exhibit reliable efficacy in diagnosing the high-risk state of significant liver fibrosis in CHB patients, and the prediction model constructed by combining them can enhance diagnostic efficacy.