Heart Attack Classification System using Neural Network Trained with Particle Swarm Optimization

TL;DR

This paper presents a neural network trained with particle swarm optimization for early heart attack detection, demonstrating superior accuracy over traditional algorithms using a novel dataset.

Contribution

It introduces a PSONN model for heart attack prediction and compares its performance against other machine learning algorithms on a new dataset.

Findings

PSONN achieved 100% accuracy, outperforming other models.

The system effectively analyzes input features for early detection.

The approach enhances prevention strategies for heart attacks.

Abstract

The prior detection of a heart attack could lead to the saving of one's life. Putting specific criteria into a system that provides an early warning of an imminent at-tack will be advantageous to a better prevention plan for an upcoming heart attack. Some studies have been conducted for this purpose, but yet the goal has not been reached to prevent a patient from getting such a disease. In this paper, Neural Network trained with Particle Swarm Optimization (PSONN) is used to analyze the input criteria and enhance heart attack anticipation. A real and novel dataset that has been recorded on the disease is used. After preprocessing the data, the features are fed into the system. As a result, the outcomes from PSONN have been evaluated against those from other algorithms. Decision Tree, Random Forest, Neural network trained with Backpropagation (BPNN), and Naive Bayes were among those employed. Then the results of 100%, 99.2424%, 99.2323%, 81.3131%, and 66.4141% are produced concerning the mentioned algorithms, which show that PSONN has recorded the highest accuracy rate among all other tested algorithms.