DLMNN Based Heart Disease Prediction with PD-SS Optimization Algorithm

Abstract

In contemporary medicine, cardiovascular disease is a major public health concern. Cardiovascular diseases are one of the leading causes of death worldwide. They are classified as vascular, ischemic, or hypertensive. Clinical information contained in patients' Electronic Health Records (EHR) enables clinicians to identify and monitor heart illness. Heart failure rates have risen dramatically in recent years as a result of changes in modern lifestyles. Heart diseases are becoming more prevalent in today's medical setting. Each year, a substantial number of people die as a result of cardiac pain. The primary cause of these deaths is the improper use of pharmaceuticals without the supervision of a physician and the late detection of diseases. To improve the efficiency of the classification algorithms, we construct a data pre-processing stage using feature selection. Experiments using unidirectional and bidirectional neural network models found that a Deep Learning Modified Neural Network (DLMNN) model combined with the Pet Dog-Smell Sensing (PD-SS) algorithm predicted the highest classification performance on the UCI Machine Learning Heart Disease dataset. The DLMNNbased PDSS achieved an accuracy of 94.21%, an F-score of 92.38%, a recall of 94.62%, and a precision of 93.86%. These results are competitive and promising for a heart disease dataset. We demonstrated that a DLMNN framework based on deep models may be used to solve the categorization problem for an unbalanced heart disease dataset. Our proposed approach can result in exceptionally accurate models that can be utilized to analyze and diagnose clinical real-world data.