A large study showed that artificial intelligence performs better than humans in the analysis of long-term monitoring of heart rhythm. Researchers from Lund University in Sweden published an article in the journal Nature Medicine, which states that the human heart beats up to 120 thousand times a day, therefore analyzing electrocardiograms, which can be recorded over days or weeks, takes a lot of time.
Using data from 14,606 patients, who wore electrocardiograms on average for 14 days, researchers initially verified data recorded by human technicians, with the help of standard methods. They then used an artificial intelligence algorithm named DeepRhythm AI, developed by the Polish company Medicalgorithmics, for secondary analysis of the data.
The researchers discovered that technicians did not notice severe arrhythmias in 4.4% of patients, whereas the artificial intelligence identified this in only 0.3% of patients. The artificial intelligence model was able to detect severe arrhythmia with a confidence level of 99.9% in the electrocardiograms that were taken over a 14-day period.
Linda Jonsson, the lead researcher from Lund University, stated: "The lack of trained personnel capable of analyzing electrocardiograms in real-time leads to significant problems in healthcare worldwide, and patients could be at risk if we continue to do shorter Holter recordings instead of longer ones." She added: "We believe that artificial intelligence can solve this problem."