Construction and Validation of a Machine Learning-Based Risk Prediction Model for Recurrent Upper Gastrointestinal Bleeding in Patients with Liver Cirrhosis: A Prospective Study

Background Upper gastrointestinal bleeding (UGIB) is a common and life-threatening complication of liver cirrhosis, with a substantial risk of recurrence despite secondary prevention. Identifying high-risk patients is essential for individualized management. Therefore, this study aimed to construct and validate a machine learning-based predictive model for UGIB recurrence in patients with liver cirrhosis, enabling accurate identification of high-risk individuals and supporting personalized management. Methods This prospective cohort study included 653 cirrhotic patients with UGIB, comprising 505 patients in the modeling cohort and 148 in the external validation cohort. Independent predictors were identified by univariate and multivariable logistic regression analyses. Four models, including logistic regression, random forest, support vector machine, and k-nearest neighbor, were developed and compared. Model performance was assessed by discrimination, calibration, and clinical utility. Results UGIB recurrence rates were 24.16% and 23.65% in the modeling and external validation cohorts, respectively. Six independent predictors were identified: age, chronic disease history, bleeding recurrence risk perception score, white blood cell count, platelet count, and serum albumin level. The logistic regression model showed the best performance, with AUCs of 0.862 (95% CI: 0.796–0.928) in internal validation and 0.827 (95% CI: 0.747–0.907) in external validation. Decision curve analysis demonstrated superior clinical benefit. Conclusion This validated model provides an accurate tool for predicting UGIB recurrence risk in cirrhotic patients and supports individualized risk stratification and management. An interactive web-based nomogram is available at https://yuceliexiantu.shinyapps.io/UGIB_Nomogram/.

Authors

Institutions

Publication Details

Journal
Research Square
Published
2026-11-18
DOI
https://doi.org/10.21203/rs.3.rs-11469264/v1
Primary Topic
Gastrointestinal Bleeding Diagnosis and Treatment
Type
preprint
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
preprint

Construction and Validation of a Machine Learning-Based Risk Prediction Model for Recurrent Upper Gastrointestinal Bleeding in Patients with Liver Cirrhosis: A Prospective Study

钟延美, Xie Hai, 邱银玲, Ziying Li et al.
5 links
Research Square
Gastrointestinal Bleeding Diagnosis and Treatment
preprint

Construction and Validation of a Machine Learning-Based Risk Prediction Model for Recurrent Upper Gastrointestinal Bleeding in Patients with Liver Cirrhosis: A Prospective Study

钟延美, Xie Hai, 邱银玲, Ziying Li, Jie Ding, Tongxia Zhan, Rui Li, Yunping Fang, Aiqin Liu
preprint en

Abstract

Background Upper gastrointestinal bleeding (UGIB) is a common and life-threatening complication of liver cirrhosis, with a substantial risk of recurrence despite secondary prevention. Identifying high-risk patients is essential for individualized management. Therefore, this study aimed to construct and validate a machine learning-based predictive model for UGIB recurrence in patients with liver cirrhosis, enabling accurate identification of high-risk individuals and supporting personalized management. Methods This prospective cohort study included 653 cirrhotic patients with UGIB, comprising 505 patients in the modeling cohort and 148 in the external validation cohort. Independent predictors were identified by univariate and multivariable logistic regression analyses. Four models, including logistic regression, random forest, support vector machine, and k-nearest neighbor, were developed and compared. Model performance was assessed by discrimination, calibration, and clinical utility. Results UGIB recurrence rates were 24.16% and 23.65% in the modeling and external validation cohorts, respectively. Six independent predictors were identified: age, chronic disease history, bleeding recurrence risk perception score, white blood cell count, platelet count, and serum albumin level. The logistic regression model showed the best performance, with AUCs of 0.862 (95% CI: 0.796–0.928) in internal validation and 0.827 (95% CI: 0.747–0.907) in external validation. Decision curve analysis demonstrated superior clinical benefit. Conclusion This validated model provides an accurate tool for predicting UGIB recurrence risk in cirrhotic patients and supports individualized risk stratification and management. An interactive web-based nomogram is available at https://yuceliexiantu.shinyapps.io/UGIB_Nomogram/.

Research Square
Weifang People's Hospital (CN), Taiwan Centers for Disease Control (TW)
Good health and well-being
Gastrointestinal Bleeding 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