Predicting spontaneous termination of atrial fibrillation using the surface ECG

Med Eng Phys. 2006 Oct;28(8):802-8. doi: 10.1016/j.medengphy.2005.11.010. Epub 2006 Jan 25.


By recognizing and characterizing conditions under which atrial fibrillation (AF) is likely to terminate spontaneously or be sustained, improved treatment of sustained AF may result and unnecessary treatment of self-terminating AF avoided. Time-frequency measures that characterize AF, such as fibrillatory frequency, amplitude, and waveform shape (exponential decay), are extracted from the residual ECG following QRST cancellation. Three complexity measures are also studied, characterizing the degree of organization of atrial activity. All measures are analysed using a training set, consisting of 20 recordings of AF with known termination properties, and a test set of 30 recordings. Spontaneous termination was best predicted by a low and stable fibrillatory frequency and a low exponential decay. Using these predictors, 90% of the test set was correctly classified into terminating and sustained AF. Neither fibrillation amplitude nor the complexity measures differed significantly between the two sets.

Publication types

  • Controlled Clinical Trial
  • Research Support, Non-U.S. Gov't

MeSH terms

  • Algorithms*
  • Artificial Intelligence*
  • Atrial Fibrillation / diagnosis*
  • Atrial Fibrillation / physiopathology*
  • Diagnosis, Computer-Assisted / methods*
  • Electrocardiography / methods*
  • Humans
  • Pattern Recognition, Automated / methods
  • Prognosis
  • Reproducibility of Results
  • Risk Assessment / methods*
  • Risk Factors
  • Sensitivity and Specificity