Atrial fibrillation is a common cardiac arrhythmia event, potentially leading to strokes and thrombosis, diagnosable by means of an electrocardiographic (ECG) exam where the patient’s heart activity is monitored continuously for several hours. The recent advances in technology have led to the development of several telemedicine applications where patients monitoring is performed in a real-time, remote fashion. Furthermore, artificial intelligence has risen as a powerful instrument for the reliable detection of heart rhythm abnor- malities, and the realization of healthcare networks would allow the prompt achievement of this task. One of the challenges in remote monitoring concerns the development of signal processing algorithms tailored to data traffic resources of an healthcare network and fitting for the involved nodes hardware/software capabilities. In this direction, we present a novel MUlti-lead Sub-beat ECG (MUSE) based technique for atrial fibrillation detection using machine learning. MUSE relies on a flexible and customizable framework, allowing the exploitation of edge computing principles to conveniently distribute the signal processing effort among different network nodes and optimize the data traffic flow as well. The proposed algorithm for atrial fibrillation detection is based on a robust principal component analysis performed on the sub-beats identified on the ECG signal coming from one or more leads. Then, the investigated signal and a subject-dependent physiological heartbeat pattern are matched to extract several metrics that drive the final ECG classification. Tests performed on public datasets and on a real Holter record demonstrate the high reliability provided by MUSE and, differently from other schemes proposed in the literature, a low sensitivity to the ECG signal quality. Moreover, the restrained computational effort required for signal processing makes MUSE perfectly tailored to the implementation in a remote healthcare network.
MUSE. MUlti-lead Sub-beat ECG for remote AI based atrial fibrillation detection / Petroni, A.; Cuomo, F.; Scarano, G.; Francia, P.; Pediconi, M.; Colonnese, S.. - In: JOURNAL OF NETWORK AND COMPUTER APPLICATIONS. - ISSN 1084-8045. - 212:(2023), pp. 1-15. [10.1016/j.jnca.2022.103544]
MUSE. MUlti-lead Sub-beat ECG for remote AI based atrial fibrillation detection
Petroni A.
;Cuomo F.;Scarano G.;Francia P.;Colonnese S.
2023
Abstract
Atrial fibrillation is a common cardiac arrhythmia event, potentially leading to strokes and thrombosis, diagnosable by means of an electrocardiographic (ECG) exam where the patient’s heart activity is monitored continuously for several hours. The recent advances in technology have led to the development of several telemedicine applications where patients monitoring is performed in a real-time, remote fashion. Furthermore, artificial intelligence has risen as a powerful instrument for the reliable detection of heart rhythm abnor- malities, and the realization of healthcare networks would allow the prompt achievement of this task. One of the challenges in remote monitoring concerns the development of signal processing algorithms tailored to data traffic resources of an healthcare network and fitting for the involved nodes hardware/software capabilities. In this direction, we present a novel MUlti-lead Sub-beat ECG (MUSE) based technique for atrial fibrillation detection using machine learning. MUSE relies on a flexible and customizable framework, allowing the exploitation of edge computing principles to conveniently distribute the signal processing effort among different network nodes and optimize the data traffic flow as well. The proposed algorithm for atrial fibrillation detection is based on a robust principal component analysis performed on the sub-beats identified on the ECG signal coming from one or more leads. Then, the investigated signal and a subject-dependent physiological heartbeat pattern are matched to extract several metrics that drive the final ECG classification. Tests performed on public datasets and on a real Holter record demonstrate the high reliability provided by MUSE and, differently from other schemes proposed in the literature, a low sensitivity to the ECG signal quality. Moreover, the restrained computational effort required for signal processing makes MUSE perfectly tailored to the implementation in a remote healthcare network.File | Dimensione | Formato | |
---|---|---|---|
Petrono_MUSE- MUlti-lead-Sub-beat ECG_2023.pdf
solo gestori archivio
Tipologia:
Versione editoriale (versione pubblicata con il layout dell'editore)
Licenza:
Tutti i diritti riservati (All rights reserved)
Dimensione
3.1 MB
Formato
Adobe PDF
|
3.1 MB | Adobe PDF | Contatta l'autore |
I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.