Body / AbdominalAI / InformaticsResearch

Logistic Regression Model Predicts Early Mortality in Cirrhotic Patients with Esophagogastric Varices

Radiology AI literature (PubMed)Jun 27

A logistic regression model predicted 28-day all-cause mortality in ICU patients with cirrhosis and esophagogastric varices with AUROC 0.882 (95% CI 0.795–0.924) in external validation, with the white cell-platelet ratio z-score as a strong determinant.

  • Retrospective study developed and validated ML models using MIMIC-IV database (n=793) with external validation at a single center (n=100).
  • LR model outperformed SVM, Naive Bayes, and Random Forest, achieving training AUROC 0.833 (0.793–0.873) and test AUROC 0.854 (0.795–0.913).
  • Single-center external validation and retrospective design limit generalizability.

Automated summary

RadPigeon summaries are original and for information only. They are not clinical advice.