Using Machine Learning Algorithms to Predict Hepatitis B Surface Antigen Seroclearance.

Area of Science:

• Hepatology
• Machine Learning in Medicine
• Virology

Background:

• Hepatitis B surface antigen (HBsAg) seroclearance indicates a better prognosis for chronic hepatitis B (CHB) patients.
• Accurate prediction of HBsAg seroclearance using clinical data remains a challenge.

Purpose of the Study:

• To identify the optimal predictive model for HBsAg seroclearance in CHB patients.
• To evaluate the performance of machine learning algorithms against traditional methods.

Main Methods:

• Utilized laboratory and demographic data from 2,235 CHB patients in the South China Hepatitis Monitoring and Administration (SCHEMA) cohort.
• Developed and compared four predictive models: extreme gradient boosting (XGBoost), random forest (RF), decision tree (DCT), and logistic regression (LR).
• Model performance was assessed using the area under the receiver operating characteristic curve (AUC).

Main Results:

• HBsAg seroclearance was observed in 106 patients.
• The XGBoost model achieved the highest AUC (0.891), outperforming RF (0.829), LR (0.680), and DCT (0.619).
• Key predictors identified by XGBoost included HBsAg level, patient age, and hepatitis B virus (HBV) DNA level.

Conclusions:

• Machine learning algorithms, especially XGBoost, demonstrate robust performance in predicting HBsAg seroclearance.
• The findings highlight the potential of leveraging machine learning with accessible clinical data for improved CHB patient management.