Risk prediction of cardiovascular disease using machine learning classifiers

Area of Science:

• Medical Informatics
• Machine Learning
• Cardiology

Background:

• Cardiovascular disease (CVD) poses significant health risks, including mortality and disability.
• Existing methods for CVD detection require improvement in performance and reliability.
• Early and automated detection systems are vital for timely intervention and improved patient outcomes.

Purpose of the Study:

• To evaluate the effectiveness of machine learning models for automatic cardiovascular disease detection.
• To compare the performance of Multi-layer Perceptron (MLP) and K-Nearest Neighbour (K-NN) algorithms in CVD classification.
• To optimize model performance through data preprocessing techniques.

Main Methods:

• Utilized publicly available University of California Irvine repository data for cardiovascular disease detection.
• Applied two machine learning techniques: Multi-layer Perceptron (MLP) and K-Nearest Neighbour (K-NN).
• Implemented data cleaning by removing outliers and attributes with null values to enhance model performance.

Main Results:

• The Multi-layer Perceptron (MLP) model achieved a detection accuracy of 82.47%.
• The MLP model demonstrated a superior area-under-the-curve (AUC) value of 86.41% compared to the K-NN model.
• Data preprocessing, including outlier and null value removal, significantly improved model performance.

Conclusions:

• The Multi-layer Perceptron (MLP) model is recommended for reliable automatic cardiovascular disease detection.
• The proposed machine learning methodology shows potential for application in detecting other diseases.
• Further validation of the MLP model using diverse datasets is suggested for broader applicability.