Enhanced Approach to Predict Early Stage Chronic Kidney Disease

The prevalence of early kidney disease (EKD) is a major public health issue that affects the well-being of patients. In order to effectively manage this illness, an early diagnosis is crucial. Here, we introduce a deep incremental learning(DIL) model that can reliably foretell when this disease would manifest in its early phases. We are testing three models: Deep Belief Network (DBN), Gated Recurrent Unit (GRU), and a new approach known as CNN+DBN. Both with and without hyperparameter adjustment are considered in this study's comparison of three models. The model achieved a 98.1% accuracy rate after training with hyperparameter tuning on a dataset of chronic renal disease. In its early stages, it shows promise as a treatment for this illness. The DBN model better handles all metrics used to evaluate model performance-accuracy, recall, F1 score, area under the curve (AUC), and precision. This research introduces a convolutional neural network (CNN) model that integrates the DBN's DIL capabilities with the CNN's feature extraction capabilities. It is capable of effectively capturing data with geographical dependencies. The model is able to process and forecast the early stages of kidney disease by utilizing the retrieved features. Combining the CNN and DBN techniques improved the accuracy of the model. It outperforms GRU+DBN models and achieves 97.15 percent accuracy with the help of hyperparameter optimization. We then proceed to process this data using the CNN+DBN model. The combined model with hyperparameter adjustment outperformed the GRU+DBN models with an accuracy of 97.15 percent. Hyperparameter adjustment also enhances the model's performance. Here lies the robustness of the applied model.