Hybrid Deep Learning Framework for Chronic Kidney Disease Prediction using 1D-CNN and Soft Voting Ensemble Classifier

Chronic Kidney Disease (CKD) is a progressive condition where early detection significantly improves patient outcomes. However, traditional diagnostic methods often struggle with the non-linear complexity and high class imbalance inherent in medical datasets. To address these challenges, this study proposes a robust hybrid framework that integrates deep learning with ensemble machine learning. The methodology employs a 1-Dimensional Convolutional Neural Network (1D-CNN) for automated feature extraction, capturing latent patterns from tabular clinical data. To mitigate data scarcity and class imbalance, the Synthetic Minority Over-sampling Technique (SMOTE) is utilized alongside controlled Gaussian noise augmentation. The extracted features are subsequently classified using a soft voting ensemble of Logistic Regression and K-Nearest Neighbors (KNN). Experimental validation on the UCI CKD dataset demonstrates that the proposed hybrid model achieves superior performance, with an accuracy of 99.6%, effectively minimizing false negatives. This approach highlights the potential of combining deep feature learning with classical ensemble stability for reliable medical diagnostics.