A chronic kidney disease prediction system based on Internet of Things using walrus optimized deep learning technique

Area of Science:

• Medical Informatics
• Artificial Intelligence in Healthcare
• Biomedical Engineering

Background:

• The Internet of Medical Things (IoMT) generates vast patient data, necessitating advanced analytical methods.
• Cloud computing (CC) and deep learning (DL) are increasingly vital for processing healthcare data.
• Accurate prediction of chronic kidney disease (CKD) is crucial for timely intervention and patient management.

Purpose of the Study:

• To propose an effective deep learning technique optimized with the walrus optimization algorithm for CKD prediction in IoMT environments.
• To enhance the accuracy and reliability of CKD prediction models by leveraging advanced feature extraction and selection methods.

Main Methods:

• Data preprocessing included missing value imputation, numerical conversion, and normalization.
• K-means (KM) clustering was used for dataset balancing to prevent prediction bias.
• Enhanced residual network 50 (EResNet50) extracted discriminative features, followed by elite opposition and Cauchy distribution-based walrus optimization algorithm (ECWOA) for optimal feature selection.
• Walrus-optimized bidirectional long short-term memory (WOBLSTM) was employed for final CKD classification.

Main Results:

• The proposed ECWOA-EResNet50-WOBLSTM model demonstrated superior performance compared to existing methods.
• Achieved a high sensitivity of 99.89% for chronic kidney disease prediction.
• The integrated approach effectively handled complex IoMT data for accurate disease prediction.

Conclusions:

• The developed deep learning technique offers a highly effective solution for CKD prediction using IoMT data.
• The optimization algorithms significantly improved feature selection and classification accuracy.
• This approach holds promise for improving early detection and management of chronic kidney disease.