Sudden cardiac death is estimated to affect ~ 400,000 people annually (1). Most of these cases are precipitated by ventricular fibrillation (VF), a chaotic abnormal electrical activation of the heart. Ventricular fibrillation disturbs systemic blood circulation and causes immediate death if therapy in the form of an electrical shock is not immediately applied. In fact, survival depends dramatically on the time it takes for therapy to arrive (2). Automated arrhythmia detection is a key component for speeding up defibrillation therapy through medical devices that detect arrhythmia and provide treatment automatically without human oversight. Examples of devices include implantable defibrillators, public access automated external defibrillators, and more.
Arrhythmias generally are an abnormal electrical activation of the heart. These abnormalities can occur in the atrial chambers, ventricular chambers, or both. Since the ventricles are the chambers responsible for providing blood to the body and lungs, disruptions in the electrical system that stimulates the mechanical contraction of the heart can be life threatening. Examples of ventricular arrhythmias include VF and ventricular tachycardia (VT), as seen in Fig. 1. Atrial arrhythmias including atrial fibrillation (AF), atrial flutter (AFl), and supraventricular tachycardia (SVT) are not immediately life threatening, but can cause uncomfortable symptoms and complications over the long term.
Figure 1. Unipolar electrograms (measurements of the electrical activity from inside the heart) for normal sinus rhythm (NSR), ventricular tachycardia (VT), and VF [10 s of the passage are shown (AAEL234) (3)].
Automated arrhythmia analysis is the detection of arrhythmias through the use of a computer. This article focuses on arrhythmia detection performed without human oversight. The primary focus will be algorithms developed for the implantable cardioverter defibrillator (ICD). An implantable cardioverter defibrillator is a device that provides an electrical shock to ventricular fibrillation and tachycardia to terminate it and restart NSR. The implantable cardioverter defibrillator was developed in the late 1970s and FDA-approved in the mid-1980s (4-7). A catheter placed in the right ventricle is used for both sensing and therapy. This device has a long history of arrhythmia detection algorithms developed in research laboratories and brought to the marketplace. Other medical devices that use purely automated arrhythmia detection include the automatic external defibrillator and the implantable atrial defibrillator. Semiautomated arrhythmia detection is used in ambulatory and bedside monitoring. These topics will be touched on briefly.
It is important to consider the measurements used to assess the performance of automated arrhythmia analysis. Sensitivity is defined as the percent correct detection of disease, while specificity is the percent correct detection of not disease. Take the case of an implantable defibrillator that detects ventricular tachycardia and ventricular fibrillation. Consider the truth table in Table 1.
Table 1. Truth Table Used to Determine Sensitivity and Specificity, Measurements of Automated Algorithm Performance
|True Positives |
|True Positives + False Negatives|
|True Negatives |
|True Negatives + False Negatives|
A false positive is one minus the specificity, while a false negative is one minus the sensitivity.