Intradialytic hypotension prediction using covariance matrix-driven whale optimizer with orthogonal...

Area of Science:

• Nephrology
• Artificial Intelligence
• Biomedical Engineering

Background:

• Intradialytic hypotension (IDH) is a serious complication during hemodialysis (HD), leading to high morbidity and mortality.
• Accurate prediction of IDH is essential for optimizing ultrafiltration prescriptions and patient safety.

Purpose of the Study:

• To develop and validate a novel prediction model (bCOWOA-KELM) for intradialytic hypotension (IDH) using routine blood test indices.
• To introduce a new optimization algorithm variant, COWOA, and its binary version (bCOWOA) for enhanced model performance.

Main Methods:

• The study proposes a novel COWOA algorithm, enhancing the Whale Optimization Algorithm (WOA) with orthogonal learning and covariance matrix adaptation.
• A binary version, bCOWOA, is used to optimize the Kernel Extreme Learning Machine (KELM) for IDH prediction.
• Model performance was validated using benchmark functions, public datasets, and a specific hemodialysis dataset.

Main Results:

• The COWOA algorithm demonstrated superior performance compared to other optimization methods across 30 benchmark functions.
• The bCOWOA-KELM model achieved a high prediction accuracy of 92.41% for IDH.
• bCOWOA outperformed other peer methods in feature selection, showing significant improvements over bSCA and bGWO.

Conclusions:

• The proposed bCOWOA-KELM model is effective for predicting intradialytic hypotension (IDH) using routine blood test data.
• The novel COWOA optimization algorithm offers significant improvements in search speed, accuracy, and convergence.
• This predictive model holds promise for future clinical applications in hemodialysis patient management.