Fully-Automated Ventricular Ectopic Beat Classification for use with Mobile Cardiac Telemetry

Benjamin Teplitzky1, Michael Mcroberts2

  • 1Preventice solutions
  • 2Preventice Solutions

Details

Category

Contributed Paper (Poster)

Sessions

18:15 - 20:15 | Mon 5 Mar | Caribbean ABC | MoPO.3

Poster Session # 1 and BSN Innovative Health Technology Demonstrations

Abstract

We have developed a robust fully-automated model for classifying ventricular ectopic beats. The classifier relies on a 12-layer convolutional neural network that was trained using over 5,000 expertly-annotated ECG recordings from the BodyGuardian® Heart (BGH) monitor. The proposed model was evaluated using data from over 950 patients and achieved classification sensitivity of 97% and specificity of 99% when evaluated using real-world ambulatory ECG recordings collected by the BGH monitor. On the MIT-BIH arrhythmia database, the classifier achieved 99% sensitivity and 99% specificity, performing equal to or better than previously published work.

Additional Information

No information added

Video

No videos found