Cardiologs, a company based in Paris, France, won FDA clearance for its software that analyzes electrocardiograms (ECG) for signs of cardiac arrhythmias. The system is based on a neural network, an artificial intelligence technique, that was trained by feeding it more than half a million previously gathered ECG recordings.
The Cardiologs system lives in the cloud and doesn’t require any equipment to invest in. It takes as input, via an online upload, long-term ECG recordings obtained from Holter monitors, ECG patches, and even smartwatches. It then carefully analyzes the data to identify suspect events that look like arrhythmias and points those out to the physician for a detailed review.
“It is intuitive that screening for AFib and subsequent anticoagulant treatment should reduce the stroke burden, which is the basis of guideline recommendations to screen for AFib in persons over the age of 65,” said Dr. Arnaud Rosier, cardiac electrophysiologist at the Hôpital Jacques Cartier, Massy in France. “Unfortunately, current R-R interval based methods to detect AFib are characterized by an inferior Positive Predictive Value (PPV) of under 59%, leading to misdiagnoses, mostly false positives, that add significant cost to the healthcare system while burdening healthcare resources and placing unnecessary stress on misdiagnosed patients or putting undiagnosed patients in harm’s way.”
More from Cardiologs about the data that led to the clearance:
When defining the reliability of diagnosing AFib and other arrhythmias, the term Positive Predictive Value (PPV) refers to the percentage of true positive cases among total cases detected. Conventional “state-of-the-art” PPV for detecting AFib is less than 59%1. The PPV for Cardiologs’ detection of AFib was 91%2 included in the cleared FDA submission. In addition, also as included in the cleared FDA submission, Cardiologs’ sensitivity for detecting AFib was reported to be 97%3 (the percentage of positive cases truly identified) and was superior to “state-of-the-art” conventional methods of detecting AFib and other arrhythmias. Cardiologs’ study results have been published in the European Journal of Preventive Cardiology (2016, Vol. 23(2S 41-55). The study’s investigators concluded: “This (Cardiologs) method may be more reliable and accurate than previous methods in the diagnosis of AFib on long-duration ambulatory ECG and other monitoring devices.”
Link: Cardiologs homepage…