Machine Learning-Based Model Used for Predicting the Risk of Hepatocellular Carcinoma in Patients with Chronic...

Area of Science:

• Hepatology
• Artificial Intelligence in Medicine
• Oncology

Background:

• Current predictive models for hepatocellular carcinoma (HCC) risk stratification in chronic hepatitis B (CHB) patients are insufficient.
• Accurate risk assessment is crucial for timely intervention and management of CHB patients at risk of developing HCC.

Purpose of the Study:

• To develop and validate an artificial neural network (ANN) model for assessing the 10-year cumulative risk of HCC in patients with CHB.
• To compare the performance of the ANN model against existing HCC risk prediction tools.

Main Methods:

• A univariate analysis identified independent risk factors for HCC development in 1717 CHB patients from two medical centers.
• An ANN model was constructed using these factors and validated on an independent cohort.
• Model performance was evaluated using area under the receiver operating characteristic curve (AUROC), concordance index (C-index), and calibration curves.

Main Results:

• The ANN model demonstrated high predictive accuracy with an AUROC of 0.929 and C-index of 0.917 in the training cohort.
• The ANN model significantly outperformed established models like mREACHE-B, mPAGE-B, HCC-RESCUE, CAMD, REAL-B, and PAGE-B (p < 0.001).
• The model effectively stratified patients into low-risk and high-risk groups with high positive and negative predictive values, confirmed in the validation cohort.

Conclusions:

• The developed ANN model shows strong performance for personalized prediction of 10-year HCC risk in CHB patients.
• This model can serve as a valuable tool for clinicians to better assess and manage HCC risk in this population.