A conceptual IoT framework based on Anova-F feature selection for chronic kidney disease detection using deep learning approach

Chronic kidney disease (CKD) is becoming an increasingly significant health issue, especially in low-income countries where access to affordable treatment is limited. Additionally, CKD is associated with various dietary factors, including liver failure, diabetes, anemia, nerve damage, inflammation, peroxidation, obesity, and other related conditions. Therefore, early prediction of CKD is important to progress the functionality of the kidney. In recent times, IoT has been widely used in a diversity of healthcare sectors through the incorporation of monitoring devices such as digital sensors and medical devices for patient monitoring from remote places. To overcome the problem, this research proposed a conceptual architecture for CKD detection. The sensor layer of the architecture includes IoT devices to collect data and the proposed classifier, MLP (Multi-Layer Perceptron), utilizes the Anova-F feature selection technique to effectively detect CKD (Chronic Kidney Disease). In addition to MLP, four other classifiers including ANN (Artificial Neural Network), Simple RNN (Recurrent Neural Network), GRU (Gated Recurrent Unit), and SVM (Support Vector Machine), are employed for comparative analysis of accuracy. Furthermore, three additional feature selection techniques, namely Chi-squared, SFFS (Sequential Floating Forward Selection), and SBFS (Sequential Backward Floating Selection), are utilized to evaluate their impact on the accuracy of CKD detection. Our proposed method outperforms all other approaches with a remarkable accuracy of 99 % while maintaining efficient computational time. This advancement is crucial in developing a highly accurate machine capable of predicting CKD in remote areas with ease.