Nutritional indicator-based prediction model for esophagogastric variceal bleeding in cirrhosis using XGBoost and SHAP.
- 1Department of Gastroenterology, First Hospital of Shanxi Medical University, Taiyuan, China.
- 2Department of Clinical Nutrition, First Hospital of Shanxi Medical University, Taiyuan, China.
Abstract
Purpose
Malnutrition is common in cirrhotic patients and associated with poor prognosis, yet current models have suboptimal predictive performance for esophagogastric variceal bleeding (EGVB). This study aimed to construct a noninvasive prediction model integrating nutritional indicators.Methods
We retrospectively analyzed 416 cirrhotic patients (135 with EGVB, 281 without EGVB) at the First Hospital of Shanxi Medical University between March 2023 and March 2025. Eight nutritional screening tools were compared by univariate logistic regression and receiver operating characteristic (ROC) analysis to identify the optimal tool. An XGBoost prediction model was then constructed with permutation-importance-based feature selection. Model performance was assessed by AUC, accuracy, sensitivity, specificity, and F1-score, with SHAP for interpretation.Results
Among the eight nutritional assessment tools, univariate logistic regression analysis revealed that MUST (OR = 1.44, 95%CI: 1.22-1.71, p < 0.001) achieved the highest predictive accuracy (AUC = 0.698). The optimal XGBoost model incorporated hemoglobin, MUST score, AST, serum creatinine, adjusted BMI, ALT, neutrophils, potassium (K⁺), monocyte count and mid-arm muscle circumference (MAMC). The model yielded an AUC of 0.860, accuracy of 84.3%, recall of 66.7%, precision of 81.8%, F1-score of 80.9%, and Brier score of 0.122. SHAP analysis demonstrated that three nutritional indicators, namely MUST score, adjusted BMI and mid-arm muscle circumference, were important predictors of EGVB. A higher MUST score, higher adjusted BMI, and lower mid-arm muscle circumference were associated with an increased risk of EGVB.Conclusion
The XGBoost-SHAP model integrating nutritional indicators accurately predicts EGVB risk in cirrhosis, outperforming single traditional nutritional screening tools. This explainable, noninvasive machine learning approach provides a reliable tool for risk stratification and individualized nutritional intervention, potentially improving outcomes in this high-risk population.
Related
- Risk factors for recurrent variceal bleeding after TIPS in cirrhotic patients with esophagogastric variceal hemorrhage: a single-center retrospective study
- The efficacy and safety of endoscopic ultrasound selective variceal devascularization for the treatment of esophagogastric varices: a retrospective cohort study.
- Platelet count combined with Child-Turcotte-Pugh score for predicting rebleeding in cirrhotic patients with esophagogastric variceal bleeding
- Can the prognostic nutritional index predict survival in patients with gastric and esophagogastric junction adenocarcinoma?
- Laboratory-based nutritional risk assessment including GDF15 in cirrhosis: a multicenter retrospective study
BibTeX
@article{zhao2026nutritional,
title = {{Nutritional indicator-based prediction model for esophagogastric variceal bleeding in cirrhosis using XGBoost and SHAP.}},
author = {Dan Zhao and Zhifang Fan and Yujie Yan and Qiaoli Guo and Qianqian Xu and Zhifen Hao and Huifang Huang},
journal = {Frontiers in nutrition},
year = {2026},
month = oct,
doi = {10.3389/fnut.2026.1911832},
url = {https://europepmc.org/article/MED/42903201},
note = {CC BY},
}
RIS
TY - JOUR
TI - Nutritional indicator-based prediction model for esophagogastric variceal bleeding in cirrhosis using XGBoost and SHAP.
AU - Dan Zhao
AU - Zhifang Fan
AU - Yujie Yan
AU - Qiaoli Guo
AU - Qianqian Xu
AU - Zhifen Hao
AU - Huifang Huang
PY - 2026
DA - 2026/10/06
JO - Frontiers in nutrition
DO - 10.3389/fnut.2026.1911832
UR - https://europepmc.org/article/MED/42903201
AB - Purpose Malnutrition is common in cirrhotic patients and associated with poor prognosis, yet current models have suboptimal predictive performance for esophagogastric variceal bleeding (EGVB). This study aimed to construct a noninvasive prediction model integrating nutritional indicators.Methods We retrospectively analyzed 416 cirrhotic patients (135 with EGVB, 281 without EGVB) at the First Hospital of Shanxi Medical University between March 2023 and March 2025. Eight nutritional screening tools were compared by univariate logistic regression and receiver operating characteristic (ROC) analysis to identify the optimal tool. An XGBoost prediction model was then constructed with permutation-importance-based feature selection. Model performance was assessed by AUC, accuracy, sensitivity, specificity, and F1-score, with SHAP for interpretation.Results Among the eight nutritional assessment tools, univariate logistic regression analysis revealed that MUST (OR = 1.44, 95%CI: 1.22-1.71, p < 0.001) achieved the highest predictive accuracy (AUC = 0.698). The optimal XGBoost model incorporated hemoglobin, MUST score, AST, serum creatinine, adjusted BMI, ALT, neutrophils, potassium (K⁺), monocyte count and mid-arm muscle circumference (MAMC). The model yielded an AUC of 0.860, accuracy of 84.3%, recall of 66.7%, precision of 81.8%, F1-score of 80.9%, and Brier score of 0.122. SHAP analysis demonstrated that three nutritional indicators, namely MUST score, adjusted BMI and mid-arm muscle circumference, were important predictors of EGVB. A higher MUST score, higher adjusted BMI, and lower mid-arm muscle circumference were associated with an increased risk of EGVB.Conclusion The XGBoost-SHAP model integrating nutritional indicators accurately predicts EGVB risk in cirrhosis, outperforming single traditional nutritional screening tools. This explainable, noninvasive machine learning approach provides a reliable tool for risk stratification and individualized nutritional intervention, potentially improving outcomes in this high-risk population.
SN - 2296-861X
ER -