AI to determine the risk of death with 15-minute ECG measurements?

There have been several reports of cases where the ECG function installed in the Apple Watch Series 4 quickly detected signs of dangerous diseases. Among these, artificial intelligence has been developed that can determine whether there is a risk of death within 30 days or within a year with only a 15-minute ECG measurement.

The CSAIL research team at MIT in the United States announced that it has developed RiskCardio, a machine learning model that can estimate the risk of patient death from an electrocardiogram. The system focuses on ACS, acute coronary syndrome, caused by poor blood flow to the heart or blockage of blood vessels. Includes cause of death, such as heart attack, myocardial infarction, and unstable angina.

The only data needed for this system is a 15-minute ECG measurement. Here, it is said that it is possible to determine whether there is a high risk of dying within 30, 60, 90 or 365 days in light of the sample heartbeat pattern. This method is based on the idea that large variability in heart rate reflects great risk. Machine learning was trained on ECG data from ACS patients. First, the heart rate pattern of each patient was classified into close groups, and then the deaths were classified as risk, and the survivors were classified as calculated, and the risk score was created by averaging them.

The advantage of Rikcardio is that it does not require additional information such as age or weight as before, and it is possible to determine the risk of death with only an electrocardiogram. Of course, this system is not yet complete. There is room for improvement, such as improving the data set in consideration of various age groups, races, and genders. However, it would certainly be useful to be able to quickly assess and determine the appropriate level of treatment if a patient has signs of risk. Related information can be found here .