Electrocardiomatrix Facilitates Accurate Detection of Atrial Fibrillation in Stroke Patients

Abstract

Background and Purpose— Cardiac telemetry is a routine part of inpatient ischemic stroke/transient ischemic attack evaluation to assess for atrial fibrillation (AF). Yet, tools to assist stroke clinicians in the evaluation of the large quantities of telemetry data are limited. The investigators developed a new method to evaluate electrocardiographic signals, electrocardiomatrix, that was applied to stroke unit telemetry data to determine its feasibility, validity, and usefulness. Electrocardiomatrix displays telemetry data in a 3-dimensional matrix that allows for more accurate and less time consuming P-wave analysis. Methods— In this single-center, prospective, observational study conducted in a stroke unit, all telemetry data from ischemic stroke and transient ischemic attack patients were collected (April 2017–January 2018) for examination facilitated by electrocardiomatrix. AF>30 seconds was identified through review of electrocardiomatrix-generated matrices by a nonphysician researcher. Electrocardiomatrix results were compared with the clinical team’s medical record documentation of AF identified through telemetry. A study cardiologist reviewed the standard telemetry associated with all AF episodes identified by electrocardiomatrix and each case of disagreement. Results— Telemetry data (median 46 hours [interquartile range: 22–90]) were analyzed among 265 unique subjects (88% ischemic stroke). Electrocardiomatrix was successfully applied in 260 (98%) of cases. The positive predictive value of electrocardiomatrix compared with the clinical documentation was 86% overall and 100% among the subset with no prior history of AF. For the 5 false-positive and 5 false-negative cases, expert overview disagreed with the clinical documentation and confirmed the electrocardiomatrix-based diagnosis. Conclusions— The application of electrocardiomatrix to stroke unit-acquired telemetry data is feasible and appears to have superior accuracy compared with traditional monitor analysis by noncardiologists.

ELEKTRA: ELEKTRokardiomatrix application to biometric identification with convolutional neural networks

“…This might be because it could be easier to identify users that all have CVD due to the dissimilarity of EKG recordings. As it is shown in other studies, it is possible to identify different cardiac conditions through the study and classification of EKGs or EKMs [165,166,125,128,140,116]. That implies that each EKM changes depending on the user if they have any CVD.…”

“…Gang Xu et. al continue their research over the EKM in [140]. In this new study, they use the EKM to detect AFIB in stroke patients.…”

“…The EKM has already been proposed in several works for medical purposes [126,127,140,128,145]. In this thesis, the main goal is to show the feasibility of EKMs for identification (cybersecurity purposes).…”

Learning Explainable Time-Morphology Patterns for Automatic Arrhythmia Classification from Short Single-Lead ECGs

“…It represents the shape and the rhythm of heartbeats in the form of a 2D matrix by splitting a given 1D ECG signal into a series of short segments of equal length containing two adjacent QRS complexes and aligning them vertically based on the first R-peak positions of the segments. In earlier studies, the ECMs were mainly used to monitor long-term ECG signals and/or detect their abnormalities by visual inspection [ 42 , 43 ].…”

Automatic Detection of Atrial Fibrillation Using Electrocardiomatrix and Convolutional Neural Network

“…This technique allows compact, two-dimensional representation of the ECG preserving morphology and rhythm characteristics, thus facilitating its review. Brown et al concluded that the ECM allows superior accuracy of manually detecting AF episodes than traditional monitor analysis by noncardiologist when using telemetry data [4]. Moreover, to improve ECG interpretation and speed up decision making for treatment, computer-based detectors of AF have shown to be helpful in diagnostic procedures to avoid further lifethreatening conditions [5].…”