Real-time wearable telecardiology from representative signals

Electrocardiograms and other similar techniques (e.g. Photoplethysmograph) are very effective tools for the detection of cardiac abnormalities. Automated analyses of ECG signals may be used for this purpose, but due to their complexity - often involving a Neural Network or Principal Component Analysis - the signal needs to be transmitted to be analysed on a powerful device. Thus, even if signals are compressed before being sent, a significant amount of non-critical information is transmitted, unnecessarily consuming bandwidth and resulting in delays. This is problematic as lives may depend on how fast and accurately an ECG signal can be analysed. We present here a fast, simple and accurate technique that works in real time to detect some ECG abnormalities. We base our analysis on correlations of the time sequence with a Representative Signal (RS) to detect abnormal behaviour. We have implemented this scheme on a standard, inexpensive portable device, so abnormalities may be automatically detected immediately on the device itself, without the need for transmission.