Development of AI for predicting coronary artery disease risk prediction of 'Stable angina'
Aug 05, 2024
The research team led by Yoon Yeon-i, Cho Young-jin, Park Ji-seok, and emergency medicine professor Kim Joong-hee at Bundang Seoul National University Hospital developed an electrocardiogram analysis artificial intelligence model that can identify high-risk groups for coronary artery disease such as myocardial infarction even in patients with 'stable angina' that do not continue chest pain unlike acute myocardial infarction.
The heart of our body receives a huge amount of blood to the heart muscle through three blood vessels called coronary arteries to repeat contraction and relaxation throughout its life, and if the coronary arteries are narrowed or blocked due to cholesterol, coronary artery disease can occur that does not properly supply blood to the heart muscle.
At this time, a condition in which the coronary artery is not blocked and the inner diameter is narrowed is called 'angina', and a disease in which blood supply is blocked due to blood clots (blood cakes) in a narrowed state and the heart muscle is paralyzed and necrotic is called 'myocardial infarction'. This includes the majority of heart diseases, which are the second leading cause of death in Korea.
When the blood supply to the heart is severely restricted due to acute myocardial infarction, the typical symptom for patients is chest pain (thoracic pain). In this case, as soon as symptoms are felt, treatment to reopen and expand blood vessels should be performed as soon as possible. Recently, artificial intelligence technology has been actively developed to determine whether acute coronary artery diseases such as myocardial infarction are present even if simple ECG is performed on chest pain patients for quick judgment and action in the emergency room.
However, most of these AI-based electrocardiogram analysis technologies can only be used for emergency patients with severe chest pain and relatively clear electrocardiogram changes, and it was difficult to find out whether there was a coronary artery problem for patients with relatively intermittent chest pain and no clear electrocardiogram change. This means that it is difficult to use for patients who do not continuously show symptoms such as chest pain when performing tests at hospitals.
Accordingly, the research team succeeded in developing an electrocardiogram analysis artificial intelligence that informs the risk of coronary artery disease in patients with stable angina by using electrocardiogram data of 21,866 patients who visited Seoul National University Bundang Hospital. At this time, coronary artery disease was defined as a narrowing of the coronary artery inner diameter by more than 50%, and the occurrence of stenosis in two or more of three blood vessels was defined as multivascular disease.
As a result of verifying the data of 4517 patients collected by the research team in a separate cohort study, the AUC (curved area), which means the accuracy of the numerical value (digital marker) calculated by the algorithm, is expected to be highly likely to be used clinically, reaching up to 0.840.
This study is meaningful as it has developed an artificial intelligence solution that can evaluate high-risk groups such as myocardial infarction in stable angina patients who have been difficult to assess the risk of coronary artery disease through electrocardiogram analysis.
Professor Yoon Yeon-i said, "The ECG results can be analyzed just by taking a picture without connecting to the electrocardiogram device, so it is a very versatile solution that anyone with a mobile phone can use. It can be widely used to select high-risk groups for coronary artery disease, not only in the emergency room but also in outpatient treatment and health checkups."
Professor Cho Young-jin has discovered new digital markers that can diagnose various diseases such as cardiovascular death, paroxysmal atrial fibrillation, left ventricular thickening, hypertrophic cardiomyopathy, and heart valve disease only with electrocardiogram""We have developed an electrocardiogram analysis solution that can be used in primary medical institutions, including them, and we plan to conduct clinical trials in the future.""
Meanwhile, the results of this study were published in the international renowned academic journal 'European Heart Journal Digital Health'.
The heart of our body receives a huge amount of blood to the heart muscle through three blood vessels called coronary arteries to repeat contraction and relaxation throughout its life, and if the coronary arteries are narrowed or blocked due to cholesterol, coronary artery disease can occur that does not properly supply blood to the heart muscle.
At this time, a condition in which the coronary artery is not blocked and the inner diameter is narrowed is called 'angina', and a disease in which blood supply is blocked due to blood clots (blood cakes) in a narrowed state and the heart muscle is paralyzed and necrotic is called 'myocardial infarction'. This includes the majority of heart diseases, which are the second leading cause of death in Korea.
When the blood supply to the heart is severely restricted due to acute myocardial infarction, the typical symptom for patients is chest pain (thoracic pain). In this case, as soon as symptoms are felt, treatment to reopen and expand blood vessels should be performed as soon as possible. Recently, artificial intelligence technology has been actively developed to determine whether acute coronary artery diseases such as myocardial infarction are present even if simple ECG is performed on chest pain patients for quick judgment and action in the emergency room.
However, most of these AI-based electrocardiogram analysis technologies can only be used for emergency patients with severe chest pain and relatively clear electrocardiogram changes, and it was difficult to find out whether there was a coronary artery problem for patients with relatively intermittent chest pain and no clear electrocardiogram change. This means that it is difficult to use for patients who do not continuously show symptoms such as chest pain when performing tests at hospitals.
Accordingly, the research team succeeded in developing an electrocardiogram analysis artificial intelligence that informs the risk of coronary artery disease in patients with stable angina by using electrocardiogram data of 21,866 patients who visited Seoul National University Bundang Hospital. At this time, coronary artery disease was defined as a narrowing of the coronary artery inner diameter by more than 50%, and the occurrence of stenosis in two or more of three blood vessels was defined as multivascular disease.
As a result of verifying the data of 4517 patients collected by the research team in a separate cohort study, the AUC (curved area), which means the accuracy of the numerical value (digital marker) calculated by the algorithm, is expected to be highly likely to be used clinically, reaching up to 0.840.
This study is meaningful as it has developed an artificial intelligence solution that can evaluate high-risk groups such as myocardial infarction in stable angina patients who have been difficult to assess the risk of coronary artery disease through electrocardiogram analysis.
Professor Yoon Yeon-i said, "The ECG results can be analyzed just by taking a picture without connecting to the electrocardiogram device, so it is a very versatile solution that anyone with a mobile phone can use. It can be widely used to select high-risk groups for coronary artery disease, not only in the emergency room but also in outpatient treatment and health checkups."
Professor Cho Young-jin has discovered new digital markers that can diagnose various diseases such as cardiovascular death, paroxysmal atrial fibrillation, left ventricular thickening, hypertrophic cardiomyopathy, and heart valve disease only with electrocardiogram""We have developed an electrocardiogram analysis solution that can be used in primary medical institutions, including them, and we plan to conduct clinical trials in the future.""
Meanwhile, the results of this study were published in the international renowned academic journal 'European Heart Journal Digital Health'.
|
bellho@sportschosun.com
