We’re extremely pleased to join the national cardiology community in applauding Dr. Mark Willcox, electrophysiologist at AHVI, for his ground-breaking study into the effectiveness of top heart monitors. As a result of his research, Dr. Willcox found that human technicians detected arrythmias twice as often as automated algorithms. As a study initially designed to compare the performance of two FDA-approved cardiac monitors—the BioGuardian by Preventice and the Carnation Ambulator Monitor (CAM) by BardyDx—Willcox’s findings were of such significance as to be presented during the Heart Rhythm Society’s annual conference.
“Not all monitors are created equal,” Dr. Willcox said during the ensuing press conference. “We all know from reading ECGs that computers aren’t quite as good as human beings, yet we rely on them a lot in the outpatient world. We are not trying to say that one monitor is clinically better than the other, because we are not comparing the clinical utility, only the diagnostic accuracy.”
Currently, Dr. Willcox’s study is the only known attempt to objectively compare the effective merits of popular heart monitor technologies. The study’s main conclusion was that the method of screening—either by an artificial intelligence (AI) program, or by a human technician—impacted the heart monitor’s performance.
Data collected by the BioGuardian monitor are first screened by an algorithm, meaning that an AI is responsible for noticing and highlighting cardiac irregularities. By contrast, the CAM monitor’s data are initially scanned by a human, and then an AI assists with recording and tracking irregularities. Willcox’s study found that CAM’s system of review—relying on human technicians to provide the first round of analysis—resulted in over a 200 percent increase in detecting critical arrhythmias.
Identifying heart arrhythmias is largely a problem of identifying patterns and weeding out legitimate risk indicators from the surrounding noise. Technicians trained to search out and investigate specific cardiac sequences are more sensitive to these warning signs. For example, a patient exhibiting a cardiac rhythm that is almost in alignment with a known arrhythmia sequence would likely be flagged by a human technician but discarded by an algorithm that can only identify a precise match to known arrhythmia sequences.
Although the Willcox study’s results indicate the superiority of human-led screenings, AI and algorithms shouldn’t be thrown out of the doctor’s office just yet. While discussing the Willcox study, Andrew Krahn, chief of cardiology at the University of British Columbia and president of the Heart Rhythm Society, pointed out an AI’s ability to quickly scan massive volumes of patient data and ongoing technology advancements as major points in AI’s favor.
Dr. Krahn noted that large-scale health data interpretation is an emerging field capable of yielding significant information for both individual and public health care. Screening individuals for disease and identifying relative severity often requires considerable data interpretation—a task that a well-designed AI can accomplish within seconds. Furthermore, the increasingly common use of ECG devices among members of the public (led by popular fitness monitors such as the Apple Watch and Fitbit) creates impossibly huge libraries of health data. An AI capable of sorting these massive data sets will therefore become an essential cardiac team member.
Willcox’s study indicates that a trained human mind can be cardiology’s most powerful diagnostic tool. With that said, relying upon those same trained minds to help build more sophisticated AI to assist with cardiac screenings is the predicted future of cardiology.
To access the full research article in press, visit https://www.heartrhythmopen.com/article/S2666-5018(21)00190-2/fulltext