Neural network has learned to predict sudden cardiac arrest

      Artificial intelligence is able to detect deadly cardiac arrhythmias in advance — 14 days before they occur. This is stated in a study published in the European Heart Journal.

      French and American scientists have analyzed more than 240,000 outpatient ECGs using a neural network that mimics the work of the human brain. The algorithm was able to identify high-risk patients with an accuracy of more than 70%. Sudden cardiac arrest kills more than 5 million people worldwide every year. The new development makes it possible to detect hidden signals of cardiac arrhythmia even before symptoms appear.

      AI was particularly effective at analyzing the temporal parameters of the cardiac cycle, from stimulation to ventricular relaxation. At the same time, he correctly identified 99.9% of patients without risk.

      The authors hope that in the future the technology will be integrated into holters and even smart watches. While the neural network is being tested, it is already paving the way for new methods of preventing sudden cardiac death.