Related Experiment Video
Updated: Apr 14, 2026
04:57
Establishing a Competing Risk Regression Nomogram Model for Survival Data
Published on: October 23, 2020
11.1K
Development and validation of machine learning nomograms for predicting mortality after cardiac valve surgery
Mateus Tamba N'dende Macho1, Yi Song2, Aojie Wei3
1Department of Cardiovascular Surgery, The First Affiliated Hospital of Zhengzhou University, Zhengzhou, China.
Frontiers in Medicine
|April 13, 2026
Summary
Machine learning models significantly improve short-term mortality prediction after heart valve surgery, outperforming the traditional EuroSCORE II. These advanced models are being developed into accessible nomograms for personalized patient risk assessment.
Area of Science:
- Cardiovascular Surgery
- Medical Informatics
- Machine Learning in Healthcare
Background:
- Heart valve surgery carries significant perioperative risks.
- Accurate mortality prediction is crucial for patient management and surgical decision-making.
- Existing risk scores like EuroSCORE II have limitations in predicting outcomes.
Purpose of the Study:
- To develop and validate machine learning (ML) models for predicting mortality after heart valve surgery.
- To compare the performance of ML models against the established EuroSCORE II.
- To create user-friendly nomograms for clinical risk stratification.
Main Methods:
- A retrospective cohort study of 935 adult patients undergoing heart valve surgery.
- Development of five models (Logistic Regression, XGBoost, Random Forest, Extra Trees, EuroSCORE II) using a 70% training set.
- Validation on a 30% hold-out set, evaluating performance using ROC AUC, sensitivity, and specificity.
Main Results:
- ML models showed strong discriminative performance for in-hospital and 30-day mortality, outperforming EuroSCORE II.
- Extra Trees (ROC AUC 0.858) and Logistic Regression (ROC AUC 0.800) were top performers for short-term mortality.
- Key predictors included age and biomarkers for cardiac stress, renal, and hepatic function.
Conclusions:
- Machine learning, especially ensemble methods, enhances short-term mortality prediction post-valve surgery.
- Developed nomograms offer a practical, interpretable tool for individualized perioperative risk assessment.
- ML-driven risk stratification can improve patient care and surgical outcomes.
