Prediction of significant liver fibrosis in a health examination population using a random forest model

BACKGROUNDLiver fibrosis is a critical stage in chronic liver disease progression, and early identification of high-risk individuals is essential.However, FibroTouch, although noninvasive, is expensive and not widely accessible for large-scale screening.AIM To develop and validate a machine learning model using routine clinical indicators to predict ≥ F2 fibrosis in health examination populations.METHODS This retrospective multicenter cross-sectional diagnostic-prediction study included a single-center cohort of 13899 participants from the First Affiliated Hospital of Zhejiang University School of Medicine (June 2021 to August 2025).The development cohort was randomly split 8:2 into training (n = 11119) and internal validation (n = 2780) sets.The tuned original model was recalibrated using a later same-hospital cohort (n = 2363; September 2025 to July 2026).Both original and recalibrated models underwent independent geographic external validation at Sichuan Provincial People's Hospital (n = 309). RESULTSSix clinical variables (body mass index, fasting insulin, alanine aminotransferase, waist circumference, triglycerides, and cholinesterase) were selected.The random forest (RF) model achieved an area under the curve (AUC) of 0.846 [95% confidence interval (CI): 0.806-0.885]and Brier score of 0.032 (95%CI: 0.028-0.037),significantly outperforming all traditional scores.Calibration was satisfactory, and decision curve analysis confirmed net clinical benefit.In the temporal cohort, the original model showed an AUC of 0.766 4 / 42 (95%CI: 0.713-0.815)and was subsequently recalibrated.In the independent geographic external validation cohort from Sichuan Province, the original and recalibrated models achieved AUCs of 0.718 (95%CI: 0.565-0.844)and 0.718 (95%CI: 0.565-0.844),respectively.Subgroup analyses showed consistent performance across age, sex, and comorbidities. CONCLUSIONThis six-indicator RF model offers an inexpensive, effective screening tool for significant liver fibrosis in Chinese individuals; however, the results should be interpreted cautiously as the reference was FibroTouch, not biopsy.

Authors

Institutions

Publication Details

Journal
World Journal of Gastroenterology
Published
2026-11-16
DOI
https://doi.org/10.3748/wjg.124330
Primary Topic
Liver Disease Diagnosis and Treatment
Type
article
Field-Weighted Citation Impact
0.00
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
article

Prediction of significant liver fibrosis in a health examination population using a random forest model

Piaopiao Jin, Liu Chen-bing, Xu Wei-Wei, Dongyu Li et al.
1 links
World Journal of Gastroenterology
Liver Disease Diagnosis and Treatment
article

Prediction of significant liver fibrosis in a health examination population using a random forest model

Piaopiao Jin, Liu Chen-bing, Xu Wei-Wei, Dongyu Li, Qi Sun, Feng Zhang, Nan Li, Zhong Liu, Chao Shen, Li-Hong Qiu
article en

Abstract

BACKGROUNDLiver fibrosis is a critical stage in chronic liver disease progression, and early identification of high-risk individuals is essential.However, FibroTouch, although noninvasive, is expensive and not widely accessible for large-scale screening.AIM To develop and validate a machine learning model using routine clinical indicators to predict ≥ F2 fibrosis in health examination populations.METHODS This retrospective multicenter cross-sectional diagnostic-prediction study included a single-center cohort of 13899 participants from the First Affiliated Hospital of Zhejiang University School of Medicine (June 2021 to August 2025).The development cohort was randomly split 8:2 into training (n = 11119) and internal validation (n = 2780) sets.The tuned original model was recalibrated using a later same-hospital cohort (n = 2363; September 2025 to July 2026).Both original and recalibrated models underwent independent geographic external validation at Sichuan Provincial People's Hospital (n = 309). RESULTSSix clinical variables (body mass index, fasting insulin, alanine aminotransferase, waist circumference, triglycerides, and cholinesterase) were selected.The random forest (RF) model achieved an area under the curve (AUC) of 0.846 [95% confidence interval (CI): 0.806-0.885]and Brier score of 0.032 (95%CI: 0.028-0.037),significantly outperforming all traditional scores.Calibration was satisfactory, and decision curve analysis confirmed net clinical benefit.In the temporal cohort, the original model showed an AUC of 0.766 4 / 42 (95%CI: 0.713-0.815)and was subsequently recalibrated.In the independent geographic external validation cohort from Sichuan Province, the original and recalibrated models achieved AUCs of 0.718 (95%CI: 0.565-0.844)and 0.718 (95%CI: 0.565-0.844),respectively.Subgroup analyses showed consistent performance across age, sex, and comorbidities. CONCLUSIONThis six-indicator RF model offers an inexpensive, effective screening tool for significant liver fibrosis in Chinese individuals; however, the results should be interpreted cautiously as the reference was FibroTouch, not biopsy.

World Journal of Gastroenterology
University Medical Center Groningen (NL), National University of Singapore (SG), National University Health System (SG), First Affiliated Hospital Zhejiang University (CN), Zhejiang University (CN)
Good health and well-being
Liver Disease Diagnosis and Treatment
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon