Survival prediction among heart patients using machine learning techniques

Area of Science:

• Medical Informatics
• Computational Biology
• Machine Learning in Healthcare

Background:

• Cardiovascular diseases (CVDs) are the leading cause of global mortality, accounting for 17.9 million deaths annually.
• The high mortality rate from CVDs necessitates advanced methods for early detection and diagnosis.
• Machine learning (ML) techniques are increasingly explored for predicting and managing cardiovascular health.

Purpose of the Study:

• To independently verify the performance of standard benchmark ML algorithms for cardiovascular disease prediction.
• To identify the most effective ML algorithm for early detection and diagnosis of heart-related diseases.
• To compare the accuracy of Decision Trees, Logistic Regression, Support Vector Machines, and Artificial Neural Networks on a curated dataset.

Main Methods:

• Utilized a standard, well-curated dataset for cardiovascular disease prediction.
• Implemented and evaluated benchmark machine learning algorithms: Decision Trees, Logistic Regression, Support Vector Machines (SVM), and Artificial Neural Networks (ANN).
• Assessed algorithm performance using a range of standard evaluation metrics.

Main Results:

• Decision Trees demonstrated superior performance compared to Logistic Regression, SVM, and ANN.
• Decision Trees achieved 14% higher accuracy than the average performance of the other evaluated algorithms.
• Contrary to some studies, Artificial Neural Networks were found to be less competitive than Decision Trees and SVMs in this evaluation.

Conclusions:

• Decision Trees are highly effective for cardiovascular disease prediction, offering significant accuracy improvements.
• The findings suggest that Decision Trees should be prioritized for developing reliable tools for early cardiovascular disease detection.
• Further research may explore ensemble methods or feature engineering to enhance ANN and SVM performance in cardiovascular risk assessment.