Nonlinear T-Wave Time Warping-Based Sensing Model for Non-Invasive Personalised Blood Potassium Monitoring in Hemodialysis Patients: A Pilot Study

Background: End-stage renal disease patients undergoing hemodialysis (ESRD-HD) therapy are highly susceptible to malignant ventricular arrhythmias caused by undetected potassium concentration ([[Formula: see text]]) variations ([Formula: see text]) out of normal ranges. Therefore, a reliable method for continuous, noninvasive monitoring of [[Formula: see text]] is crucial. The morphology of the T-wave in the electrocardiogram (ECG) reflects [Formula: see text] and two time-warping-based T-wave morphological parameters, [Formula: see text] and its heart-rate corrected version [Formula: see text] , have been shown to reliably track [Formula: see text] from the ECG. The aim of this study is to derive polynomial models relating [Formula: see text] and [Formula: see text] with [Formula: see text] , and to test their ability to reliably sense and quantify [Formula: see text] values. Methods: 48-hour Holter ECGs and [[Formula: see text]] values from six blood samples were collected from 29 ESRD-HD patients. For every patient, [Formula: see text] and [Formula: see text] were computed, and linear, quadratic, and cubic fitting models were derived from them. Then, Spearman’s ([Formula: see text]) and Pearson’s (r) correlation coefficients, and the estimation error ([Formula: see text]) between [Formula: see text] and the corresponding model-estimated values ([Formula: see text]) were calculated. Results and Discussions: Nonlinear models were the most suitable for [Formula: see text] estimation, rendering higher Pearson’s correlation (median 0.77 [Formula: see text] 0.92) and smaller estimation error (median 0.20 [Formula: see text] 0.43) than the linear model (median 0.76 [Formula: see text] 0.86 and 0.30 [Formula: see text] 0.40), even if similar Spearman’s [Formula: see text] were found across models (median 0.77 [Formula: see text] 0.83). Conclusion: Results support the use of nonlinear T-wave-based models as [Formula: see text] sensors in ESRD-HD patients.