Cardiac Failure Forecasting Based on Clinical Data Using a Lightweight Machine Learning Metamodel.

Area of Science:

• Cardiology
• Artificial Intelligence
• Health Informatics

Background:

• Heart failure is a critical condition with multiple contributing risk factors.
• Machine learning (ML) offers potential for predicting heart disease but faces implementation challenges.
• Accurate prediction is vital for preventing life-threatening cardiac events.

Purpose of the Study:

• To develop and evaluate a machine learning metamodel for predicting heart failure.
• To assess the predictive performance of the proposed metamodel using combined clinical datasets.
• To compare the metamodel's accuracy against other ML models for heart failure prediction.

Main Methods:

• A machine learning metamodel was constructed using Random Forest Classifier, Gaussian Naive Bayes, Decision Tree, and k-Nearest Neighbor algorithms.
• The metamodel was trained and validated on a consolidated dataset from five distinct heart disease datasets.
• Eleven standard clinical features were utilized across all datasets for model development.

Main Results:

• The proposed machine learning metamodel demonstrated a high prediction accuracy of 87% for heart failure.
• The metamodel outperformed other individual machine learning models in predicting heart failure.
• The study highlights the efficacy of ensemble methods in cardiovascular risk prediction.

Conclusions:

• The developed machine learning metamodel provides an accurate method for heart failure prediction.
• This approach holds significant potential for clinical utility in early diagnosis and patient management.
• Further research can explore integrating more diverse data sources to enhance predictive capabilities.