The Apple Watch compatible sensor AliveCor KardiaBand detects dangerous potassium levels in blood with 94% accuracy. Nowadays, such tests usually use needles for invasive blood tests. KardiaBand, produced by AliveCor, allows the user to touch the sensor and then read the heart's electrical activity called an electrocardiogram (ECG). Vic Gundotra, CEO of AliveCor, presented the findings to the Mayo Clinic, claiming that the same technology can detect high levels of potassium in the blood, called hyperkalemia.
Hyperkalemia may be caused by diabetes, dehydration, and chronic kidney disease. It does not cause kidney and heart failure. Too much potassium interferes with the electrical activity of cells, including heart cells. According to Gundotra, the ECG mode can show too much potassium. AliveCor partnered with the Mayo Clinic to develop a new CardiaBand algorithm that can analyze ECG data and detect whether the user has hyperkalemia. The data set includes 2 million ECGs, which are associated with 4 million K values, which were collected over 23 years.
In order to use these data points to train artificial intelligence, the team divided the data set into several parts. They use some data to train the network. Basically, they have read electrocardiogram patterns to show hyperkalemia and let AI learn how to find patterns. After the training was completed, the team tested the artificial intelligence on different parts of the data, achieving an accuracy rate of 90% to 94%.