Nonlinear T-Wave Time Warping-Based Sensing Model for Non-Invasive Personalised Blood Potassium Monitoring in...

Area of Science:

• Cardiology
• Nephrology
• Biomedical Engineering

Background:

• End-stage renal disease patients on hemodialysis are prone to dangerous heart arrhythmias due to potassium fluctuations.
• Continuous, noninvasive monitoring of potassium levels is essential for patient safety.
• T-wave morphology in ECGs correlates with potassium changes, offering a potential monitoring avenue.

Purpose of the Study:

• To develop polynomial models relating T-wave morphological parameters (dw, dw,c) to potassium concentration changes (Δ[K+]).
• To evaluate the accuracy and reliability of these models in sensing and quantifying Δ[K+] in hemodialysis patients.

Main Methods:

• Collected 48-hour Holter ECGs and blood potassium levels from 29 hemodialysis patients.
• Computed T-wave morphological parameters (dw, dw,c) and derived linear, quadratic, and cubic models.
• Assessed model performance using Spearman's (ρ) and Pearson's (r) correlation coefficients and estimation error (ed).

Main Results:

• Nonlinear models demonstrated superior performance for Δ[K+] estimation compared to linear models.
• Higher Pearson's correlation coefficients (median 0.77–0.92) and lower estimation errors (median 0.20–0.43) were observed with nonlinear models.
• Spearman's ρ values were similar across all models, indicating robust rank correlation.

Conclusions:

• Nonlinear T-wave-based models are effective for estimating potassium concentration changes in hemodialysis patients.
• These models show promise as noninvasive sensors for real-time potassium monitoring.
• The findings support the clinical utility of ECG-derived parameters for managing ESRD-HD patients.