Related Experiment Video
Updated: Jul 4, 2025
Human Liver Microphysiological System for Assessing Drug-Induced Liver Toxicity In Vitro
Published on: January 31, 2022
LiveBoost: A GB-based prediction system for liver fibrosis in chronic hepatitis B patients in China - A multi-center
Guoxiang Xie1, Huanming Xiao2, Quan Liu1
1Human Metabolomics Institute Inc., Shenzhen, Guangdong, China.
Background:
The aim of this study was to evaluate the accuracy of LiveBoost™, a gradient boosting (GB)-based prediction system based on standard biochemical values (AST, ALT, platelet count) and age, in Chinese patients with chronic hepatitis B (CHB) and compare its performance with FIB-4 (fibrosis-4 score) and APRI (the aspartate transaminase to platelet ratio index).
Methods:
This retrospective trial enrolled 454 participants, including 279 CHB patients who underwent liver biopsy and 175 normal controls from 3 centers in China. All participants underwent laboratory blood testing. LiveBoost was constructed using GB and FIB-4 and APRI were calculated from laboratory data.
Results:
LiveBoost outperformed APRI and FIB-4 in predicting hepatic fibrosis and cirrhosis. The GB model had an AUROC of 0.977 for CHB diagnosis, 0.804 for early and advanced fibrosis, and 0.836 for non-cirrhosis and cirrhosis, compared to AUROC of 0.554, 0.673 and 0.720 for FIB-4, AUROC of 0.977, 0.652 and 0.654 for APRI.
Conclusions:
LiveBoost is a more reliable and cost-effective method than APRI and FIB-4 for assessing liver fibrosis in Chinese patients with CHB.
