Machine learning-based prediction of 1-year mortality in hypertensive patients undergoing coronary revascularization

Amir Hossein Behnoush1,2,3,4, Amirmohammad Khalaji1,2,3,4, Malihe Rezaee1,2,4,5

  • 1Tehran Heart Center, Cardiovascular Diseases Research Institute, Tehran University of Medical Sciences, Tehran, Iran.

Clinical Cardiology
|January 2, 2023
PubMed

Insights

Machine learning models accurately predict 1-year mortality in hypertensive patients after coronary artery bypass graft (CABG) surgery. Logistic regression demonstrated the highest predictive accuracy, aiding risk assessment for high-risk subgroups.

Area of Science:

  • Cardiology
  • Medical Informatics
  • Machine Learning

Background:

  • Hypertensive patients face elevated mortality risks post-coronary artery bypass graft (CABG).
  • Machine learning (ML) shows potential in improving medical predictions, including preventive cardiology.
  • Accurate mortality prediction is crucial for managing high-risk patient groups.

Purpose of the Study:

  • To develop and assess five machine learning models for predicting 1-year mortality in hypertensive patients undergoing CABG.
  • To identify key predictors of mortality in this patient population.
  • To compare the performance of different ML algorithms for this specific clinical application.

Main Methods:

  • Utilized Tehran Heart Center's CABG registry data, including baseline and peri-procedural characteristics.
  • Employed random forest (RF) for feature selection, identifying 11 significant predictors.
  • Developed and evaluated five ML models: logistic regression (LR), RF, artificial neural network (ANN), extreme gradient boosting (XGB), and naïve Bayes (NB), using AUC, sensitivity, and specificity.

Main Results:

  • The study included 8,493 hypertensive patients, with 303 deaths within the first year.
  • Total ventilation hours and ejection fraction were identified as leading predictors of mortality.
  • Logistic regression (LR) achieved the highest predictive performance with an AUC of 0.82, outperforming other models.

Conclusions:

  • All developed ML models demonstrated excellent performance in predicting 1-year mortality for hypertensive CABG patients.
  • Logistic regression (LR) emerged as the superior model based on AUC.
  • These ML tools can assist clinicians in assessing mortality risk, particularly in high-risk subgroups like hypertensive patients.
Abstract