An evaluation of RiskCardio, from MIT’s Computer Science and Artificial Intelligence Laboratory, demonstrates how the system’s high-risk patients – patients in the top quartile – were nearly seven times more likely to die of cardiovascular death when compared to the low-risk group in the bottom quartile. By comparison, patients identified as high risk by standard metrics were only three times more likely to suffer an adverse event.
The technology analyzes electrocardiogram (ECG) signals (with no additional medical data required). The machine learning algorithm constructs a framework to learn from biometric signals. This is especially valuable where where patient-level labels are available but signal segments are rarely annotated.
The technology focuses on patients who have survived an acute coronary syndrome (ACS). This refers to a set of syndromes which occur due to decreased blood flow in the coronary arteries such that part of the heart muscle is unable to function properly or dies.
From the study, the researchers were able to reframe risk stratification for cardiovascular death as a multiple instance learning problem. In addition, the scientists were able to demonstrate how the framework can be used to design a new risk score, for which patients in the highest quartile are 15.9 times more likely to die of cardiovascular death within 90 days of hospital admission for an acute coronary syndrome. Understanding this enables preventative actions to be taken.
According to lead researcher, Dr. Divya Shanmugam: “We’re looking at the data problem of how we can incorporate very long time series into risk scores, and the clinical problem of how we can help doctors identify patients at high risk after an acute coronary event.”
The research has been published as a white paper, titled “Multiple Instance Learning for ECG Risk Stratification.”
