Comparative Analysis of the Effectiveness of Riskometer Scales in Predicting the Risk of in-Hospital Mortality in

B I Geltser1, K I Shahgeldyan2, I G Domzhalov1

  • 1Far-East Federal University, School of Medicine, Ajax Bay, Russky Island.

Kardiologiia
|September 12, 2024
PubMed

Insights

The GRACE, CADILLAC, and TIMI-STe scales effectively predict in-hospital death (IHD) in ST-segment elevation myocardial infarction (STEMI) patients post-PCI. An integrated artificial neural network (ANN) model demonstrated superior accuracy in identifying high-risk patients.

Area of Science:

  • Cardiology
  • Medical Informatics
  • Biostatistics

Background:

  • ST-segment elevation myocardial infarction (STEMI) is a critical condition requiring timely intervention.
  • Percutaneous coronary intervention (PCI) is a primary treatment for STEMI.
  • Accurate prediction of in-hospital death (IHD) is crucial for managing STEMI patients post-PCI.

Purpose of the Study:

  • To comparatively evaluate existing riskometer scales for predicting IHD in STEMI patients after PCI.
  • To develop and assess novel machine learning-based models for improved IHD prediction.
  • To stratify risk accurately for STEMI patients undergoing PCI.

Main Methods:

  • Retrospective analysis of 4,675 STEMI patient records undergoing emergency PCI.
  • Evaluation of GRACE, CADILLAC, TIMI-STe, PAMI, and RECORD scales for IHD risk stratification.
  • Development of prognostic models using logistic regression, gradient boosting, and artificial neural networks (ANN).

Main Results:

  • The GRACE scale exhibited the highest prognostic accuracy; PAMI showed the lowest.
  • CADILLAC and TIMI-STe scales demonstrated acceptable prognostic abilities.
  • The integrative ANN model, incorporating scores from five scales, outperformed logistic regression and gradient boosting in prediction accuracy.
  • ANN model-based stratification identified low, medium, high, and very high IHD risk groups.

Conclusions:

  • GRACE, CADILLAC, and TIMI-STe scales are advantageous for IHD risk stratification in STEMI patients post-PCI.
  • The integrated ANN model offers superior prediction quality and accurate identification of high-risk STEMI patients.
  • Machine learning models, particularly ANN, enhance the accuracy of IHD risk prediction in STEMI patients post-PCI.
Abstract