Chronic kidney disease detection using XceptionNet with Harmonic Addax Optimization

Area of Science:

• Nephrology and Artificial Intelligence
• Medical Imaging and Diagnostics
• Computational Health Sciences

Background:

• Chronic Kidney Disease (CKD) is a progressive kidney function impairment leading to serious health complications.
• Early CKD detection is crucial for preventing kidney failure and managing secondary conditions like hypertension and diabetes.
• Existing detection methods face challenges with generalization and class imbalance, hindering accurate early diagnosis.

Purpose of the Study:

• To develop and validate an optimized deep learning framework for enhanced early detection of Chronic Kidney Disease (CKD).
• To address limitations in current CKD detection methods, specifically generalization and class imbalance issues.
• To improve the accuracy and reliability of CKD diagnosis for better patient outcomes.

Main Methods:

• A novel CKD detection framework integrating XceptionNet with the Harmonic Addax Optimization Algorithm (HAOA) was proposed.
• The framework employed sigmoid normalization, feature fusion using a Deep Belief Network (DBN) with a Soergel metric, and Synthetic Minority Overlapping Technique (SMOTE) for data augmentation.
• The HAOA algorithm was developed by combining Harmonic analysis and the Addax Optimization Algorithm (AOA).

Main Results:

• The proposed Xception with HAOA model demonstrated high performance across multiple CKD datasets.
• Achieved a True Positive Rate (TPR) of 94.679%, True Negative Rate (TNR) of 92.777%, and an overall accuracy of 93.667%.
• Reported a precision of 92.258% and an F1-score of 93.453%, indicating robust diagnostic capabilities.

Conclusions:

• The developed deep learning framework offers an effective solution for early CKD diagnosis.
• The integration of XceptionNet and HAOA significantly improves detection accuracy and overcomes existing limitations.
• This approach has the potential to reduce kidney failure rates and enhance patient prognosis through timely intervention.