Artificial intelligence (AI) could help interpret electrocardiogram (ECG) results, supporting healthcare staff in the diagnosis of diseases that affect the heart.
Scientists from Uppsala University and heart specialists in Brazil have designed an AI that is capable of diagnosing atrial fibrillation and five other general ECG abnormalities automatically, just like a cardiologist.
The research has been reported in the Nature Communications journal.
An ECG is a simple test done to check the electrical activity and rhythm of the heart. The test results are displayed on a graph and can disclose several conditions that affect the heart. Hence, the tool has been utilized regularly in healthcare, and results of each ECG must be interpreted by a cardiologist manually.
The recent study illustrates that an AI can automatically diagnose the abnormalities pointed out by an ECG. Initially, the AI was trained on a database including more than two million ECGs that had earlier been diagnosed manually.
Thus, the AI can learn to identify normal patterns for the six most common ECG abnormalities and further make a diagnosis of another patient with one such condition—with similar accuracy as that of a cardiologist.
Currently, the technique is not ready to be used in hospitals and clinics. But the scientists hope that AI provides huge potential for enhanced cardiovascular care in low- and middle-income countries where a large part of the population lacks the same access to experts who can interpret ECG results as in Sweden.
This is the first result of a collaboration that we have built up over the past two years. I have great confidence that in the future this type of deep collaboration between AI researchers and medical researchers will be able to create new knowledge that can help people enjoy a better quality of life.
Thomas Schön, Professor of Automatic Control, Uppsala University
Schön works in AI and machine learning at Uppsala University and is also responsible for the study’s technical part.
The study is based on a mathematical model (called deep artificial neural network), which is a good example of the fundamental concept behind machine learning, in which computers develop their own model and use it to learn to solve problems based on the gathered information.
The technique varies from the standard method of working with a computer, wherein the computer has been programmed manually to carry out a highly specific task.
The results obtained for several problems have been proved better upon using machine learning, and the computer itself can recognize patterns from collected texts, diagrams, images, and figures.
Ribeiro, A. H., et al (2020) Automatic diagnosis of the 12-lead ECG using a deep neural network. Nature Communications. doi.org/10.1038/s41467-020-15432-4.