Healthcare is a basic human need. Cardiovascular disorders are sometimes referred to as “heart diseases.” Early methods of assessing what should have been done for those at high risk for cardiovascular disease may have helped such people. The major goal of identifying and processing and generated from based on readings of the heart detection of irregularities in heart conditions, which can save human lives. This article's goal aims to use deep learning to build a hybrid model methodology using CNN and Bi-LSTM to predict whether one who suffers from heart disease and use that prediction to raise awareness or make a diagnosis. We apply data processing methods to heart disease Cleveland uci dataset which addresses both missing data and data imbalance and is available to the public. While CNN-Bi-LSTM was used for classification, an auxiliary tree classifier was used for feature selection. Many tests have been conducted. Conducted using UCI Heart Disease Dataset from Cleveland Clinic that was gathered through Kaggle. Several Classification, accuracy, precision, recall, and the f1-score are only few of the measures used. Used to evaluate a diagnostic model's performance.