Detecting cardiovascular disease (CVD) early to improve outcomes for patients. This study presents a new method for predicting cardiac disease using U-NET architecture for accurate cardiac segmentation and repeated neural networks (RNNs) for continuous data analysis. To effectively predict heart disease, RNN models are used to examine temporal trends in patient data, including clinical records and ECG signals. U-NET is used using improved characteristic extraction to separate the morphology of medical images, such as MRI and CT scans, containing accurate images of heart structure. The combination of these two models uses the advantages of both architectures. U-NET provides accurate spatial analysis, while RNN provides dynamic temporal prediction. The effectiveness of the binding approach in increasing the identification accuracy of heart disease has been demonstrated by experimental results, and its potential as a powerful tool for its potential clinical decisions. The proposed method is more effectively than standard procedures and may provide an exciting way forward for modern cardiovascular treatment.