AI-DRIVEN CLOUD HEALTHCARE SYSTEM FOR PREDICTIVE ANALYSIS OF CARDIOVASCULAR DISEASES

Abstract

Cardiovascular disease (CVD) is still one of the world's main reasons of mortality, imposing considerable burdens on healthcare systems and economies, and hence requiring timely and precise risk prediction for enhanced patient outcomes. The existing CVD risk assessment approaches, such as statistical models and traditional machine learning methods, have limitations like poor scalability, poor management of heterogeneous data, poor handling of timely processing, and weaknesses in data privacy, which hamper their seamless integration into clinical workflows. To bridge these loopholes, this study proposes a hybrid AI framework combining Deep Neural Networks (DNN) and XGBoost on a cloud-native platform that enables smooth processing of big data without causing non-compliance with security expectations. The originality of this approach is that it is a fusion of ensemble learning and deep learning within a cloud-based secure environment which allows for real-time model updating and individualized risk estimation. The experimental results determine greater predictive ability, 99.95% precision, F1 measure of 98.99%, and highly balanced values of recall and precision, considerably surpassing baseline models such as individual XGBoost, Random Forest and CNN-LSTM hybrids. The hybrid architecture demonstrates greater robustness, less false negative and higher clinical utility compared to existing methods. This advancement significantly enhances the feasibility of early diagnosis and facilitates scalability for deployment with flexibility across a variety of healthcare applications. Future research attempts to further minimize false negatives, include multimodal data sources, and enhance model interpretability.