Introduction: Integrating machine learning algorithms into the medical field has become essential for improving disease diagnosis and predicting conditions at an early stage. However, conventional machine learning techniques often struggle with large, complex datasets, limiting their effectiveness. This work proposes a novel deep-learning approach to enhance disease prediction accuracy using medical databases to address this. Methods: This study introduces a two-stage deep learning model utilizing a Convolutional Neural Network (CNN) for disease prediction. CNNs, known for their strengths in pattern recognition and regression, are applied to classify medical data. In the first stage, initial classification is performed, while the second stage focuses on analysing the experimental dataset to assess accuracy. To evaluate the performance of the proposed CNN model, a comparative study is conducted against two established models: VGG16 and Recurrent Neural Network (RNN). Results: The proposed CNN model achieved an accuracy of 98.5%, significantly surpassing the performance of both VGG16 (85%) and RNN (90%). The CNN's ability to handle complex datasets with diverse medical parameters effectively highlights its superiority in disease prediction tasks. Conclusion: The study demonstrates that the CNN-based deep learning model offers a highly accurate and efficient solution for early disease prediction, outperforming traditional models. With its improved accuracy and robustness, the proposed approach has the potential to enhance diagnostic capabilities, enabling timely medical interventions and better patient care outcomes.