| Abstract [eng] |
Electrolytes play an essential role in the regulation of the heart, and the ECG can be a useful instrument for detecting metabolic disturbances. Potassium is important in several physiological processes, including in the maintenance of the resting cellular-membrane potential and the propagation of action potential for the excitability in cardiac tissue. Hyperkalaemia, usually clinically silent, is a medical condition known for higher serum potassium levels and it is one of the causes of sudden cardiac deaths of chronic kidney disease patients. These patients, with residual kidney function, rely exclusively on haemodialysis to eliminate toxic wastes and to maintain overall homeostasis. In order to develop a nonobtrusive method capable of monitoring potassium changes remotely, this study investigated which ECG parameters are sensible to fluctuations during the interdialytic period. Specifically, the study evaluated a hypothesis that new model-based ECG features are able to discriminate the changes of electrolytes. Short term ECGs (1-minute-long) were acquired from 33 patients immediately before and after haemodialysis using a multiparametric body scale. During the study, an algorithm was developed to denoise the signals and extract the principal features. The algorithm is capable of detecting the T wave, J point and QRS complex, and examining the T wave polarity. Only positive T waves were considered during this study. The last stage of the algorithm consists in modeling the T wave and ST segment onto a Gaussian and Lognormal function. The algorithm was implemented in MATLAB environment. Results show that T wave features: right slope, left slope, center of gravity, and lognormal fitting parameters show statistically significant differences if evaluated before and after haemodialysis. Therefore, it could be concluded that the proposed algorithm could be useful in developing of unobtrusive telemetry system for monitoring of hemodialysis patients at home. |
