AI-based Apple Watch app detects arrhythmia with 97% accuracy
An AI-based Apple Watch app has been shown to detect heart arrhythmias with 97% accuracy, according to new study data.
As reported by MobiHealthNews, the results come from an ongoing investigation called the mRhythm Study which is researching the use of technology in detecting heart health issues.
The primary focus of the study is the Cardiogram Apple Watch App - an app that uses a deep learning algorithm to collect clinically relevant and actionable heart rate data.
The algorithm was created through the enrolment of 6,158 Cardiogram users. Heart health data, such as heart rate and mobile electrocardiogram insights, was used to train the algorithm before it was subjected to a cohort of 51 patients undergoing a procedure called cardioversion - a procedure where electricity or medication is used to normalised an abnormally fast heart rate.
The Cardiogram app was used both before and after the procedure via an Apple Watch, specifically to identify atrial fibrillation (an abnormal heart rate). The condition currently affects more than 2.7 million US citizens.
The app correctly identified atrial fibrillation with a 97% accuracy, 98% sensitivity and 90.2% specificity.
“Our results show that common wearable trackers like smartwatches present a novel opportunity to monitor, capture and prompt medical therapy for atrial fibrillation without any active effort from patients,” said the study's senior author Dr Gregory Marcus. “While mobile technology screening won’t replace more conventional monitoring methods, it has the potential to successfully screen those at an increased risk and lower the number of undiagnosed cases of AF.”
The study is the latest example of the successful use of an AI algorithm in healthcare. Researchers at Google recently created an algorithm capable of detecting breast cancer from digitised pathology slides to a higher degree of sensitivity than trained pathologists, whilst an algorithm created by researchers at the University of Texas at Austin accurately predicted major depressive disorder in 75% of cases.
mRhythm Study researchers will now continue to train the algorithm, possibly in a wider range of conditions.