Stanford researchers claim to have developed an algorithm that “exceeds the performance of board certified cardiologists in detecting a wide range of heart arrhythmias from electrocardiograms [ECG] recorded with a single-lead wearable monitor,” according to a study published in arXiv.
The team used the Zio patch from iRhythm Technologies, a San Francisco, CA startup, which allowed them to gather ECG recordings over a period of up to two weeks. These recordings were run against a computer running a deep learning algorithm that was trained by analyzing almost 30,000 previously gathered and diagnostically assessed ECG recordings. The result is that the system is now able to spot 14 different types of cardiac arrhythmias purportedly better than the six Stanford cardiologists that were pitted against it.
Here’s a Stanford video with the researchers that developed the algorithm:
Related flashbacks: Medgadget Exclusive Interview with iRhythm’s CEO Kevin King…; ZIO Wireless Patch May Be Better Option Than Holter Monitors for Cardiac Arrhythmia Diagnosis…;
Study in arXiv: Cardiologist-Level Arrhythmia Detection with Convolutional Neural Networks…