Efficient Associated Feature Vector Using Two Level Recurrent Neural Network For Cardiovascular Disease Classification

Abstract

Diseases of the heart and blood vessels called Cardiovascular Diseases (CVDs) are among the most common and potentially fatal conditions affecting people today. Early diagnosis of heart disease via symptoms is a big challenge in the current medical system. Therefore, there is a need for a less expensive and less invasive technique that can detect heart diseases. Correctly diagnosing heart disease in advance allows doctors to more effectively treat patients in the interim, reducing the likelihood of a cardiac patient experiencing a heart attack. The goal is achieved by training a deep learning model on a large data set pertaining to healthcare delivery for cardiovascular disorders. Multiple recently established deep learning-based solutions for both cardiac disease prediction and diagnosis are available however several limitations exists. Lacking a sophisticated framework, systems struggle to process high-dimensional datasets for usage in a wide variety of data sources for cardiac disease prediction. Several variations of Recurrent Neural Network (RNN) models have been created for sequenced features, and this model type lends itself readily to temporal sequenced data. RNN uses three distinct activation functions during categorization. Given a large enough training set, recurrent neural networks are able to accurately predict a future diagnosis of heart failure. The size of the training set appears to have a linear relationship with the model's performance. The proposed model designs an Efficient Associated Feature Vector using Two Level Recurrent Neural Network (EAFV-TLRNN) for accurate classification of CVDs. The proposed model performance is compared with the traditional models and the results represent that the proposed model performance is high.