Performance Comparison of Random Forest Classifier and Convolution Neural Network in Predicting Heart Diseases

Abstract

The applications of machine learning (ML) in the digital era become inevitable. Few domains include virtual personal assistants, predictions while commuting, audio and video surveillance, filtering of email spam and malware(s), service and support in online and social media, refining the search engine performance, online fraud detection, product recommendations, healthcare, finance, travel, retail, media, and so on. Among the various functionalities, the applications of ML in the health domain play a momentous role. The objective of the paper is to focus the applications of ML in predicting the cardiac arrest/heart attack based on the earlier health records. Though there exists opulence of data on the history regarding the cardiac diseases, the inadequacy in analyzing and predicting the heart attack leads to sacrifice the human life. The research focuses on predicting the cardiac arrest/heart attack using the ML approaches based on the patient’s historical data. Among the various ML techniques, the paper focuses on random forest classifier (RFC) and convolution neural network (CNN)-based prediction methods. The experimentation was conducted on the standard datasets available in the UCI repository. The results concluded that RFC had outperformed the other classifier regarding the classification accuracy.