Related Experiment Videos

Prediction of 1-year mortality after heart transplantation using machine learning approaches: A single-center study

Ying Zhou1, Si Chen1, Zhenqi Rao1

  • 1Department of Cardiovascular Surgery, Union Hospital, Tongji Medical College, Huazhong University of Science and Technology, Wuhan, China.

Insights

Machine learning models predict heart transplant (HTx) survival. The Random Forest model shows high accuracy, identifying high-risk patients for personalized care and reducing organ waste.

Area of Science:

  • Cardiology
  • Medical Informatics
  • Machine Learning

Background:

  • Heart transplantation (HTx) is the standard treatment for end-stage heart failure.
  • Accurate prognosis assessment is crucial for optimizing patient outcomes.
  • Machine learning offers novel approaches for predictive modeling in healthcare.

Purpose of the Study:

  • To develop and validate a machine learning-based risk prediction model for 1-year mortality after heart transplantation.
  • To identify key predictors of post-transplant mortality.
  • To enhance personalized treatment strategies for HTx recipients.

Main Methods:

  • Utilized data from 381 consecutive orthotopic heart transplant recipients (2015-2018).
  • Employed Least Absolute Shrinkage and Selection Operator (LASSO) for variable selection.
  • Developed and compared seven machine learning models, including Random Forest (RF) and Gradient Boosting Machine (GBM), with bootstrap validation.
  • Applied Shapley Additive exPlanations (SHAP) for model interpretability.

Main Results:

  • Albumin, recipient age, and left atrium diameter were the most significant predictors of 1-year mortality.
  • The Random Forest (RF) model achieved the highest discrimination, with an Area Under the Curve (AUC) of 0.801.
  • Gradient Boosting Machine (GBM) demonstrated the best sensitivity (0.271).
  • SHAP analysis provided individual-level insights into RF model predictions.

Conclusions:

  • A validated machine learning risk-prediction model for heart transplantation prognosis was successfully established.
  • The RF model exhibits superior predictive performance for 1-year mortality.
  • This model can aid in identifying high-risk patients, tailoring therapies, and minimizing organ wastage.
Abstract

Related Concept Videos