High blood potassium levels constitute a condition known as hyperkalemia. It can be related to a number of causes, including kidney disease, dehydration, injury and diabetes and hyperkalemia can affect heartbeat rhythm. Yesterday, during the American College of Cardiology conference, AliveCor presented work done with the Mayo Clinic showing that its technology can detect hyperkalemia through EKGs.
The researchers used electrocardiogram data collected from 709,000 patients over the course of 23 years, which included 2.1 million EKGs and 4 million blood potassium measurements. Two-thirds of that data were used to train a neural network to detect hyperkalemia through EKG readings. That network was then tested on the other third of the data. It was able to correctly detect cases of hyperkalemia 85 percent of the time and accurately label those that weren't hyperkalemic 72 percent of the time. The team also tested it on 10 patients using an experimental AliveCor smartphone EKG device and from almost 50 hours of data -- 5.4 of which recorded a hyperkalemic state -- the neural network was able to correctly spot hyperkalemia 94 percent of the time and non-hyperkalemia 74 percent of the time.
It's important to note that these findings are preliminary and AliveCor's technology hasn't been approved by the FDA to detect hyperkalemia. The company is continuing with its trials, The Verge reports, and will likely seek FDA clearance in the future.