NON-UNIFORM DISCRETE ECG REPRESENTATION OPTIMISED FOR MEDICAL DATA FIDELITY

A review of optimal ECG data representation is motivated by spreading networks of home care monitoring, emerging as a valuable tool for interactive monitoring in home care. Most of currently available devices use full disclosure signal transmission what rises the cost of monitoring or alert message transmission relying on the interpretation intelligence built in a wearable recorder. Our approach consists in the automatic on-line record evaluation aiming at extracting the local density of medical information followed by re-sampling of gathered data stream to a non-uniform representation. The correlation of medical data distribution with ECG waves was confirmed in the study of scanpath density function in context of detected wave borders. For the implementation of ECGdedicated data optimization algorithm we apply continuous non-uniform sampling providing best compromise of data density and quality. For the CSE signals the average compression ratio was 3,01 and the distortion level within QRS was 0,22% (3,3 μV).