2019/01/11
Keio University School of Medicine
A research group from the Division of Cardiology at the Keio University School of Medicine, led by Associate Professor Motoaki Sano and Assistant Professor Shinichi Goto, has developed a technology using artificial intelligence (AI). This technology can instantly determine from a single electrocardiogram whether a patient visiting the emergency room with chest pain requires catheterization, with an accuracy of over 80%.
The heart is made of a muscle called the myocardium and acts as a pump to send blood throughout the body. In the blood vessels (coronary arteries) that supply the oxygen and nutrients the heart needs to function, deposits of lipids (plaque), primarily "bad cholesterol," build up with age. Acute coronary syndrome is a condition that occurs when this plaque ruptures, causing a blood clot (thrombus) to suddenly form in a coronary artery, which severely restricts or completely blocks blood flow.
Among these, acute myocardial infarction is a condition in which myocardial necrosis progresses rapidly, and if left untreated, it can lead to life-threatening situations such as pump failure, lethal arrhythmia, or cardiac rupture. To treat this, it is crucial to perform catheterization as quickly as possible on the blocked vessel to restore coronary blood flow. To shorten the time from onset to the start of catheterization, a rapid diagnosis is essential. The electrocardiogram plays a central role in diagnosing acute coronary syndrome, but the diagnostic ability of physicians varies greatly, and the accuracy of the automated ECG analysis currently installed in electrocardiographs is not satisfactory. On the other hand, catheterization, which involves inserting a thin tube from an artery in a limb to near the heart, is an invasive procedure with risks and cannot be performed indiscriminately on all patients complaining of chest pain. To accurately identify patients who truly need catheterization, the development of more accurate automated ECG analysis was needed.
This technology was developed by training the AI on electrocardiograms, using "whether emergency catheterization was actually performed to widen the coronary arteries" as the correct answer. When a new electrocardiogram is provided, it instantly determines the need for catheterization. In other words, what experienced cardiologists previously determined by comprehensively evaluating various information, including catheterization results, can now be replicated by this technology with an accuracy of over 80% using just a single electrocardiogram. An electrocardiogram can be performed immediately upon a patient's visit, with results available within minutes. By incorporating this technology into electrocardiographs, it becomes possible to automatically, quickly, and accurately diagnose the need for catheterization and suggest it to physicians. It is expected to reduce deaths from heart disease, the second leading cause of death in Japan, and contribute to extending healthy life expectancy.
The results of this research were published in the online edition of the American scientific journal "PLOS ONE" on January 9, 2019, at 2:00 p.m. (US Eastern Time).
The full press release can be found below.