Related Experiment Video
Updated: Aug 27, 2025
11:34
A Competent Hepatocyte Model Examining Hepatitis B Virus Entry through Sodium Taurocholate Cotransporting Polypeptide as a Therapeutic Target
Published on: May 10, 2022
2.3K
Using machine learning models to predict HBeAg seroconversion in CHB patients receiving pegylated interferon-α
Hongyan Shang1,2, Yuhai Hu3, Hongyan Guo4
1Department of Laboratory Medicine, Gene Diagnosis Research Center, The First Affiliated Hospital of Fujian Medical University, Fuzhou, China.
Journal of Clinical Laboratory Analysis
|October 1, 2022
Summary
Predicting treatment success in chronic hepatitis B (CHB) is crucial. An XGBoost model using routine lab data effectively predicts HBeAg seroconversion in CHB patients receiving pegylated interferon-alfa (PegIFN-α) therapy.
Area of Science:
- Hepatology
- Virology
- Machine Learning in Medicine
Background:
- Chronic hepatitis B (CHB) affects millions globally.
- Pegylated interferon-alfa (PegIFN-α) offers treatment benefits but has limited response rates (30-40%).
- Identifying predictors for PegIFN-α response is essential to improve patient outcomes.
Purpose of the Study:
- To explore baseline predictors for HBeAg seroconversion in CHB patients.
- To establish and validate predictive models for PegIFN-α treatment response.
- To enhance the efficacy of PegIFN-α therapy through personalized treatment strategies.
Main Methods:
- A cohort of 260 HBeAg-positive CHB patients receiving PegIFN-α monotherapy was randomly divided into training (70%) and testing (30%) sets.
- Feature selection was performed using Recursive Feature Elimination, Boruta, and LASSO algorithms on 50 routine laboratory variables.
- Eight machine learning models, including XGBoost, were trained and evaluated for predicting HBeAg seroconversion.
Main Results:
- The XGBoost model demonstrated superior performance with high AUROC values (0.900 in training, 0.910 in testing).
- XGBoost showed excellent calibration, accurately predicting HBeAg seroconversion.
- Key predictors identified by XGBoost included treatment duration, HBV DNA (log), HBeAg, HBeAb, HBcAb, ALT, triglyceride, and ALP levels.
Conclusions:
- The XGBoost model, utilizing common laboratory variables, is a robust tool for predicting HBeAg seroconversion in CHB patients on PegIFN-α.
- This predictive model can aid clinicians in optimizing PegIFN-α treatment strategies.
- Further research can refine these models for broader clinical application in managing CHB.
