Overlapped Symptoms Detection for Cardiovascular Disease Based on Deep Learning Model

Cardiovascular diseases (CVD) have a significant impact on increasing the mortality rate in the Middle East has one of the highest age-standardized death rates for cardiovascular disease (CVD). Recently, based on the Assessment Risk Tools for Cardiovascular Diseases (CVD), World Health Organization (WHO) reported that 40% of all fatalities are attributed to cardiovascular diseases which has been linked to the main Risk Factors (RF) advances as obesity, hypertension, tobacco, and high cholesterol. In most of the cases, angiography is a reliable method for the diagnosis and treatment of cardiovascular diseases. However, it is a costly approach associated with various complications. The significant increase in the prevalence of cardiovascular diseases and the subsequent complications and treatment costs have urged researchers to plan for the better examination, prevention, early detection, and effective treatment of these conditions. The present study aimed to determine the patterns of cardiovascular diseases using integrated classification techniques for analyzing the data of internal medicine patients who are at the risk of heart failure with 2621 samples and 40 characteristics. Selecting the characteristics and evaluating the influential factors are essential to the development of classifiers and increasing their accuracy. The proposed work suggested a model based on Gini-Entropy-Regression Model (GERM). The objective is to predict future risk with a certain probability and compared its performance with Deep Learning MLP Model. Statistical analysis and methods were used in this research to detect the symptoms that overlapped and to accurately identify a specific heart condition. The dataset utilized to train the computer consists of medical records from 14 hospitals which were collected based on four main categories such as basic information, symptoms, inducement and history, and physical sign and assistant examination. The suggested model consisted of four levels, level 1: Preprocessing data, Level 2: Feature Extraction, Level 3: Feature Selection, Level 4: Feature Detection. The results of the suggested model were as follows: the result was 84.4% when the symptoms of (CVD) is overlapping DSYP and CHEP. When Accuracy measured with combination DSYP, CHEP, and CYAN it has been increased up to 88.9%. DSYP, CHEP, CYAN, showing values of 89.8%. in 5th Neural Network (NN) the combinations were DSYP, CHEP, CYAN, DBPH, WFAT, EMPT showing ideal value of accuracy measured up to 90.6% and with Fever this combination of Neural Network has been showing accuracy = 91%. From the findings the previous seven predictors (Risk Factors) giving the best overlapping and diagnosis for CVD.