Chronic Kidney Disease Prediction using CNN, LSTM and ensemble model

Abstract

Chronic Kidney Disease (CKD) is a serious and progressive disease that impacts kidney function over time, more likely than not resulting in kidney failure unless caught early. In this research, we suggest a predictive model for CKD based on deep learning strategies, such as Convolutional Neural Networks (CNN), Long Short-Term Memory (LSTM) networks, and an ensemble model that integrates both methods. The CNN model learns spatial patterns from clinical data, and the LSTM model learns temporal relationships, which makes the system reliable for sequential clinical records. The ensemble model combines the merits of both architectures to increase the accuracy of predictions and decrease false positives. Experimental results show that our hybrid system is better than classical machine learning models with respect to precision, recall, and F1-score. This research underscores the promise of early CKD detection through deep learning, which will facilitate early medical intervention and better patient outcomes.