Prediction of risk factors and electrocardiographic changes in chronic kidney disease patients

Area of Science:

• Nephrology
• Cardiology
• Biomedical Engineering

Background:

• Chronic kidney disease (CKD) presents a significant global health challenge, increasing morbidity and mortality, especially from cardiovascular events.
• Effective management of CKD necessitates early identification and intervention for risk factors to prevent disease progression and associated complications.
• The heterogeneity of CKD across diverse populations complicates the development of generalized predictive strategies.

Purpose of the Study:

• To develop accurate predictive models for cardiovascular events in patients diagnosed with chronic kidney disease.
• To integrate physiological parameters captured by biosensors with machine learning techniques for enhanced predictive capabilities.
• To explore the utility of genetic and non-genetic risk scores in conjunction with physiological data for cardiovascular risk assessment in CKD.

Main Methods:

• Utilized biosensors to capture key physiological parameters: Oxygen saturation (SpO₂), Pulse rate (PR), Perfusion index (Pi), Respiration rate (RRp), and Pleth variability index (PVi).
• Applied Stacked Auto-Encoders (SAEs) for diagnostic analysis of the captured physiological data.
• Developed Genetic Risk Score (GRS) and Non-Genetic Risk Score (NGRS) models using natural logarithms of odds ratios (OR) for risk factor evaluation.

Main Results:

• Developed predictive models integrating weighted contributions of various factors to forecast cardiovascular events in CKD patients.
• A novel machine learning approach incorporating automatic machine learning (AutoML) was implemented to enhance model accuracy.
• The integrated models demonstrated potential for improved cardiovascular risk prediction in the CKD population.

Conclusions:

• The developed models effectively integrate physiological properties and risk factors with weighted contributions for predicting cardiovascular events in CKD.
• The incorporation of automatic machine learning (AutoML) represents a significant advancement in predictive modeling for CKD complications.
• These findings support the use of advanced machine learning and comprehensive physiological monitoring for proactive cardiovascular risk management in CKD patients.