Artificial Intelligence Accurately Detects Heart Failure in a Second
Data on one heartbeat is enough for neural networks to identify a dangerous condition. The authors of the development claim that you don’t even need to go to the clinic - the development can be integrated into smartwatches. About 10% of people over 65 suffer from congestive heart failure, as a result, the heart loses its ability to effectively pump blood through the body. To identify this disease doctors use an electrocardiogram and to make a diagnosis, measurements are carried out for a couple of minutes, sometimes it takes several days in a row. Researchers at the University of Surrey have proposed a new approach - much simpler and faster, and also more reliable. They developed a convolutional neural network that analyzes ECG results and identifies the disease in just a few seconds. Large public databases with ECG results from healthy people and patients with heart failure were used to train the algorithm. According to the creators, the neural network can diagnose the disease with absolute accuracy, based on data on one heartbeat. However, the technology has a number of limitations. The fact is that only severe cases of heart failure were used to test it. This means that milder forms of this disease may not be recognized. Therefore, before the neural network begins to be used by doctors massively, thorough clinical trials should be conducted.