Neural Network Approaches for Real-Time Detection of Cardiovascular Abnormalities SEEJPHVolume XXV,S2, 2024, ISSN: 2197-5248;Posted:05-12-2024 2283 | Page Neural Network Approaches for Real-Time Detection of Cardiovascular Abnormalities Kiran Kumar Maguluri 1 , Chandrashekar Pandugula 2 , Zakera Yasmeen 3 1 IT systems Architect, Cigna Plano Texas, ORCID: 0009-0006-9371-058X 2 Sr Data Engineer, Lowes Inc NC, USA, ORCID: 0009-0003-6963-559X 3 Data engineering lead Microsoft, ORCID: 0009-0004-8130-2111 Dr. Aaluri Seenu, Professor, Department of CSE, SVECW, Bhimavaram, AP, India KEYWORDS Early Detection, Cardiovascular Diseases, Real-Time Detection, Deep Learning Models, Complex Models, Convolutional Neural Networks, Time-Frequency Information, Discrete Wavelet Transform, Cardiac Diseases, Model Optimization, Health Monitoring, Machine Learning, Predictive Models, Real-Time Use, Computational Efficiency, Signal Processing, Disease Detection, Neural Networks, Medical Imaging, Model Analysis. ABSTRACT The early detection of cardiovascular diseases could be life- saving, especially when the location of the patient is considered. Therefore, in recent years, work has been done on the early detection of cardiovascular diseases. The common point of deep learning models developed for the detection of cardiovascular diseases is the use of complex models. The complex model not only increases the amount of calculations but also prevents real- time use for the detection of cardiac diseases. In this study, by using simple deep learning models, the aim is to determine the deep learning model that allows the real-time detection of cardiovascular diseases. For this purpose, in the study, the models developed using convolutional neural networks and time- frequency information obtained with discrete wavelet transform were analyzed. 1. Introduction Almost one third of deaths that occur each year in advanced countries are caused by cardiovascular disease. A major concern in these cases is the length of time that is required to evaluate patients’ clinical conditions after an intermediate-to-severe cardiovascular abnormality is detected using current technologies. Automatic analysis of the daily life-sustaining signals that are used to control these diseases can address this issue if combined with deep learning models to interpret the patterns of such signals. With these features, cyber-physical systems can revolutionize clinical healthcare. This paper presents an ensemble of viable deep neural networks that were used to create an automatic real-time abnormality detection system through which several cardiovascular abnormalities are well treated or avoided. The creation of such a system is motivated by the need to evaluate and treat patients with cardiovascular abnormalities more rapidly. This need is based on two different facts: the shortage of cardiologists that is currently experienced at the Brazilian National Health System and the high temporal correlation that is observed between the responses of a non-invasive, robust, and physical subject-independent heart sound biomarker to the early appearance of cardiovascular