• Back to the workflow
  • Preliminary Content
    • Acknowledgements
    • Preface
    • Dedication
    • Abstract
  • Introduction
  • 1 Objectives and the research question
  • 2 Related Works
  • 3 The planned approach and methods for solving the problem
    • 3.1 State of the art
    • 3.2 Research plan and methods
      • 3.2.1 Type of study
      • 3.2.2 The data
      • 3.2.3 Workflow
      • 3.2.4 Statistical analysis
      • 3.2.5 Research Team
    • 3.3 Expected results and outcomes
  • 4 Whatever
  • References

Detecting life-threatening patterns in Point-of-care ECG using efficient memory and processor power

2 Related Works

The CinC/Physionet Challenge 2015 produced several papers aiming to reduce false alarms on their dataset. On Table it is listed the five life-threatening alarms present in their dataset.

Table 2.1: Definition of the 5 alarm types used in CinC/Physionet Challenge 2015 challenge.
Alarm Definition
Asystole No QRS for at least 4 seconds
Extreme Bradycardia Heart rate lower than 40 bpm for 5 consecutive beats
Extreme Tachycardia Heart rate higher than 140 bpm for 17 consecutive beats
Ventricular Tachycardia 5 or more ventricular beats with heart rate higher than 100 bpm
Ventricular Flutter/Fibrillation Fibrillatory, flutter, or oscillatory waveform for at least 4 seconds

They used as score the following formula, which penalizes five times the false negatives (since we do not want to miss any real event):

\[Score=\frac{TP+TN}{TP+TN+FP+5*FN}\]

The five-best scores in this challenge are presented on Table 10–14.

Table 2.2: Challenge Results on Streaming
Score Authors
81.39 Filip Plesinger, Petr Klimes, Josef Halamek, Pavel Jurak
79.44 Vignesh Kalidas
79.02 Paula Couto, Ruben Ramalho, Rui Rodrigues
76.11 Sibylle Fallet, Sasan Yazdani, Jean-Marc Vesin
75.55 Christoph Hoog Antink, Steffen Leonhardt

Their algorithm did a pretty good job on the Physionet test-set. However, independently of their approach to this problem, none of the authors reported benchmarks, memory usage, robustness test, or context invariance that could assure its implementation on real monitors to reduce alarm fatigue indeed.

There are other arrhythmias that this challenge did not assess, like atrial standstill (hyperkalemia), third-degree atrioventricular block, and others that may be life-threatening in some settings. Pulseless electrical activity is a frequent condition in cardiac arrest but cannot be identified without blood pressure information. This information is usually present in ICU settings but not in other locations.