Cardiologist-level arrhythmia detection and classification in ambulatory electrocardiograms using a deep neural network

Authors: Awni Y. Hannun, Pranav Rajpurkar, Masoumeh Haghpanahi, Geoffrey H. Tison, Codie Bourn, Mintu P. Turakhia and Andrew Y. Ng.

Summary:

A large, single-lead ECG study developed and validated a deep neural network that classified 12 rhythm types from Zio patch recordings with cardiologist-level performance suggesting AI can accurately triage and interpret ambulatory ECGs.

Key Findings:

Zio deep neural network AI validated against board certified cardiologists.

Zio F score (harmonic mean of the positive predictive value and sensitivity) exceeded that of average cariologists (0.780).

The content on iRhythm Academy is provided for general educational purposes only and should not be considered the exclusive source for this type of information. This training does not replace or supersede approved labeling. The content will be shared with physicians and healthcare professionals who seek a deeper understanding of the operation and use of iRhythm products and services with the intent of enhancing their knowledge of features and operations described in the clinician manuals. At all times, it is the professional responsibility of the physicians and healthcare professionals to exercise independent clinical judgment in a particular situation. The device and service functionality and programming described on iRhythm Academy are based on its products and services, and can be referenced in the labeling and reference manuals. This website is intended for healthcare professionals and not intended for patients or consumers.