The functionality of the Human++ demonstrator(*) "wireless ECG patch for cardiac monitoring" has been extended with software that delineates the electrocardiography (ECG) signal locally on the patch node before sending the results over the air to the receiver. The algorithm achieves excellent results for sensitivity and predictivity, and covers a broad range of wave morphologies.

The wireless ECG patch is a hybrid system combining electronic assembly on a flexible polyimide substrate and integration in textile. This enables flexibility in one dimension and stretchability in the other, which is required for optimal personal comfort. The patch features IMEC's proprietary ultra-low-power biopotential ASIC, a commercial microcontroller and a 2.4GHz radio link. The patch can continuously monitor the patient's heart at a sample rate of up to 1kHz. It sends the results directly to the receiver, or it can delineate the signals locally before sending them. Local delineation reduces the use of the radio, improving the autonomy of the patch. The current autonomy with local delineation is 10 days of continuous monitoring.

For local delineation, a wavelet-based ECG delineator algorithm has now been implemented. The algorithm transforms the monitored ECG signal using discrete wavelet transform, and then performs a multi-scale search for the ECG waves. The delineator is able to identify P,Q, R, S, and T wave peaks and boundaries. Because the intervals and amplitudes of these waves contain most of the useful information of the ECG, this delineation will provide quick and useful information to the healthcare provider.

The QRS-wave search uses scales 21 to 23. It is based on modulus maximum lines, and the decision whether to consider a maximum modulus will be made immediately. The search will continue with adjacent slopes to delineate further QRS waves. The algorithm covers QRS, QR, RS, R, and QS wave morphologies. The search for P and T waves also uses scales 23 to 24. It follows the QRS delineation and covers the following morphologies: positive and negative, raising and falling waves for both P and T waves, and biphasic T waves.

The delineator on the ECG patch has been validated over all the records in the MIT-QT database. It achieves a 99.93% sensitivity and a 98.28% positive predictivity for QRS detection on 86,994 beats. For delineation over 3,623 beats, it reaches a 99.83% sensitivity and a 95.08% positive predictivity.

(*) Human++ is the Holst Centre strategic program in which technology innovations in the domain of health and wellness monitoring are demonstrated and tested.

Back