AArD Algorithm Performance Evaluation

<< Back to all published studies

Performance of the Advanced Arrhythmia Discrimination Algorithm During Wearable Cardioverter Defibrillator Use


  • The purpose of this analysis was to study the impact of a noise processing algorithm (Advanced Arrhythmia Discrimination Algorithm) on the arrhythmia alarm rate experienced in patients using the LifeVest WCD.

About the Advanced Arrhythmia Discrimination Algorithm (AArD)

  • The AArD algorithm further discriminates noise in ECG signals by extracting and analyzing features, including intensity and frequency, beyond the previous discrimination algorithm employed by the LifeVest WCD. This algorithm was developed with supervised machine learning methods using an ECG dataset that is independent from the data used for analysis in the current publication.
  • AArD was introduced in 2018 and became standard in all LifeVest devices by 2019.

Key Results

  • AArD produced a 56% relative reduction in total arrhythmia alarms compared to already low rates seen in historical performance.
  • The benefit described here is related to reductions in alarms caused by artifact sensing; the initial discrimination algorithm and the newer AArD algorithm have little impact on the over detection of actual arrhythmias.
  • Median # of false detections over 90 days with AArD was 0 compared to a median of 4 without (p = <0.001).
    • The majority of LifeVest patients had 0 false alarms over 90 days with AArD.
  • The AArD algorithm did not prolong the time to appropriate shock delivery for the population; the median time was 44 seconds with AArD compared to 45 seconds without (p=ns).


  • Analysis demonstrates that the majority of LifeVest patients experience few false alarms.
  • The reduction in false alarms with use of the AArD algorithm did not affect efficacy of the WCD; there was no safety impact related to delays in appropriate shock delivery with the AArD algorithm.
  • Data published before the advent of the AArD algorithm have reported arrhythmia alarm rates that, while historically accurate, do not reflect device enhancements and are no longer relevant to current clinical practice.

Background and Methods

  • Retrospective comparative study of 96,000 randomly selected patients from the years 2017 and 2019.
    • Random samples of 4,000 patients/month over the entire 12 months from each of the years 2017 and 2019 were taken to control for seasonal effects.
  • Comparisons between calendar years 2017 and 2019 was performed using the chi-square test for categorical data and the Wilcoxon test for continuous data.
  • Included patients were prescribed the LifeVest WCD during the years 2017 (Discrimination Algorithm, DA group) or during 2019 (Advanced Arrhythmia Discrimination Algorithm, AArD) who resided in the United States.

Click here to read the full publication
Arkles, J., Delaughter, C. & D’Souza, B. A novel artificial intelligence based algorithm to reduce wearable cardioverter-defibrillator alarms. J Interv Card Electrophysiol (2023). https://doi.org/10.1007/s10840-023-01497-w  

<< Back to all published studies