Predicting Coronary Heart Disease using Various Regression Analysis

Abstract

Nowadays, cardiovascular diseases are the major concern for Human beings. Cardiovascular Heart Diseases (CHDs) are the major reason for mortality globally causing millions of deaths each year. The global death toll from Cardiovascular Heart Disease (CHD) rose from 12.1 million in 1990 to 20.5 million in 2021. The age-standardized death rate for CHD in India was 282 deaths/100000 was higher compared with global levels that is 233 deaths/100000. The leading factors contributing to cardiovascular diseases or fatalities encompass high cholesterol level, triglyceride, etc. These are the most common factors nowadays which result in heart attack in humans. The main aim of this particular project is to explore various risk factors associated with myocardial infarction, commonly known as Heart Attack. We have used tensorflow and Keras frameworks for the following project. In this project, we utilize neural network models to forecast the risk of coronary heart disease (CHD) using various features. The prediction process involves training the models on a portion of the data and assessing their effectiveness on the remaining test set. These models are designed to discern patterns and correlations between input features and the target variable, 'chd'. The choice of features, architecture, and training parameters influences the model's predictive performance. The research focuses on two distinct neural network models: the first, 'sbp_model,' predicts CHD using only systolic blood pressure ('sbp'), while the second, 'linear_model,' utilizes all available features after normalization. Both models are evaluated on their ability to predict CHD through mean absolute error, with training histories and loss curves analysed. We have taken into consideration all the important Regression Models.