An efficient cardiovascular disease prediction and risk analysis evaluation using deep learning techniques with attention mechanisms

Abstract

Cardiovascular disease (also known as CVD) continues to be the newlinelargest cause of death across the world. Because of its complicated nature and newlinehigh mortality rates, cardiovascular illness presents substantial difficulties to newlinehealthcare systems. Real-time electrocardiogram (ECG) signal improvement newlineand predictive modeling are the primary focuses of this thesis, which provides newlinea newlinecomplete strategy to improving the prediction and diagnosis of newlinecardiovascular disease (CVD) through the development of sophisticated newlinemachine learning and signal processing techniques. newlineInitial phase of the research contribution is done by developing an newlinealgorithm known as Online Recursive Independent Component Analysis newline(ORICA), which is specifically designed to remove artifacts from newlineelectrocardiogram data in real time. The constraints of the conventional newlineIndependent Component Analysis (ICA) are addressed by this innovative newlinetechnique, which enables the mixing matrix to be continuously updated as newlinenew data is received. The ORICA approach shows a considerable increase in newlinethe clarity of the electrocardiogram (ECG) signal by efficiently reducing noise newlineand artifacts, such as baseline drift and muscle noise, which are typical in newlineclinical settings. An improved interpretation of cardiac events was achieved newlineas a consequence of the improved signal quality, which led to an increase in newlinethe reliability of diagnoses based on electrocardiograms. newline