The use of a wearable ECG that monitors the heart continuously results in hundreds of hours of data that must be inspected second by second for any indications of problematic arrhythmias. Some of these symptoms are extremely difficult to differentiate from harmless heartbeat irregularities.
This data management problem was tackled by Stanford University researchers in collaboration with heartbeat monitor company iRhythm. A massive dataset was amassed to train a deep neural network model, which, after seven months, was able to diagnose arrhythmias about as accurately as cardiologists and outperform them in most cases. The group took approximately 30,000, 30-second clips from various patients that represented a variety of arrhythmias to develop a system that detects 14 types of arrhythmia.
Six different cardiologists were individually assigned to diagnose the same 300-clip set. The researchers then compared which ones closely matched the consensus opinion — the algorithm or the cardiologists working independently. The algorithm was shown to be competitive with the cardiologists and outperformed them on most arrhythmias.
And, unlike its human counterparts, the algorithm does not experience fatigue and can make arrhythmia detections instantaneously and continuously.
Long term, the group hopes this algorithm could be a step toward expert-level arrhythmia diagnosis for people who don’t have access to a cardiologist, as in many parts of the developing world and in other rural areas. More immediately, the algorithm could be part of a wearable device that at-risk people keep on at all times that would alert emergency services to potentially deadly heartbeat irregularities as they’re happening.