Abstract
World Health Organization reports that cardiovascular disease is the major cause of death. Cardiovascular disease is responsible for around 31% of global deaths, resulting in over 17.9 million deaths, making it a global health issue. Classifiers and common methods for encoding categorical data using machine learning have revealed a broad variety of surprising findings when used to identify cardiac disease based on small datasets from testing. To extract features without developing a comprehension of sequence information, the early research used convolutional neural networks (CNN)—a type of deep learning model. A deep learning-based system, particularly a CNN with bidirectional long/short-term memory, is proposed in this study in order to efficiently predict cardiovascular disease from patient data. Only the most applicable features were selected via feature selection, which was accomplished by prioritizing and picking features that were highly rated in the supplied disease dataset. After that, the CNN + BiLSTM-based hybrid deep learning technique was employed to predict cardiovascular illness. In contrast to past studies of a similar kind, the experimental outcomes of this hybrid deep learning technique were encouraging: 94.507% accuracy, % precision, % recall, and an F1-score of 94%.