A Hybrid Machine Learning Approach for Cardiac Arrest Prediction Using Ensemble Techniques

Abstract

Cardiac arrest remains one of the leading causes of mortality worldwide, often occurring without prior warning. Early prediction and detection can significantly improve survival rates and enable timely medical intervention. This project proposes a hybrid machine learning-based system for predicting cardiac arrest using multiple classification techniques integrated into a web-based platform built with Django. The system utilizes patient-related attributes such as age, sex, chest pain type, resting blood pressure, ECG results, maximum heart rate, exercise-induced angina, and other clinical indicators. These features are extracted from a dataset and preprocessed to generate meaningful inputs for machine learning models. A key aspect of the system is the transformation of input data using CountVectorizer, which converts textual identifiers into numerical vectors suitable for training. Multiple machine learning algorithms are employed, including Artificial Neural Networks (MLPClassifier), Support Vector Machines (SVM), Logistic Regression, and Decision Tree Classifier. These models are trained individually and evaluated based on performance metrics such as accuracy, confusion matrix, and classification report. To enhance prediction reliability, an ensemble learning technique using a Voting Classifier is implemented. This hybrid model combines predictions from all individual classifiers to produce a more robust and accurate output. The system is deployed as a web application where users can register, log in, input medical parameters, and receive predictions indicating the presence or absence of cardiac arrest risk. Additionally, the system stores prediction results and provides analytical insights such as prediction ratios and accuracy comparisons across models. The proposed system demonstrates improved predictive performance by leveraging ensemble learning and offers a user-friendly interface for real-time prediction. It serves as a decisionsupport tool for healthcare professionals and individuals, contributing to early diagnosis and preventive healthcare strategies.