Development and validation of a novel fusion algorithm for continuous, accurate, and automated R-wave detection and calculation of signal-derived metrics

J Crit Care. 2013 Oct;28(5):885.e9-18. doi: 10.1016/j.jcrc.2013.02.015. Epub 2013 Apr 22.

Abstract

Purpose: Previous studies have shown that heart rate complexity may be a useful indicator of patient status in the critical care environment but will require continuous, accurate, and automated R-wave detection (RWD) in the electrocardiogram (ECG). Although numerous RWD algorithms exist, accurate detection remains a challenge. The purpose of this study was to develop and validate a novel fusion algorithm (Automated Electrocardiogram Selection of Peaks, or AESOP) that combines the strengths of several well-known algorithms to provide a more reliable real-time solution to the RWD problem.

Materials and methods: This study involved the ECGs of 108 prehospital patient records and 32 ECGs from a conscious sedated porcine model of hemorrhagic shock. The criterion standard for validation was manual verification of R waves.

Results: For 108 human ECG records, the AESOP algorithm overall outperformed each of its component algorithms. In addition, for 32 swine ECG records, AESOP achieved an R-wave sensitivity of 97.9% and a positive predictive value of 97.5%, again outperforming its component algorithms.

Conclusion: By fusing several best algorithms, AESOP uses the strengths of each algorithm to perform more robustly and reliably in real time. The AESOP algorithm will be integrated into a real-time heart rate complexity software program for decision support and triage in critically ill patients.

Keywords: Automatic data processing; Clinical decision support systems; Electrocardiography; Heart rate complexity; Signal detection analysis.

Publication types

  • Research Support, U.S. Gov't, Non-P.H.S.
  • Validation Study

MeSH terms

  • Algorithms*
  • Animals
  • Critical Illness*
  • Electrocardiography
  • Heart Rate / physiology*
  • Humans
  • Predictive Value of Tests
  • Shock, Hemorrhagic / physiopathology*
  • Signal Processing, Computer-Assisted*
  • Swine