Scientists from MIT Computer Science & Artificial Intelligence Lab presented an AI model RiskCardio to estimate risk of cardiovascular death.
RiskCardio measures the electrical activity of a patient’s heart and based on this data predicts the possibility of cardiovascular death. According to the researchers, only first 15 minutes of a patient’s raw electrocardiogram (ECG) signal is enough for the system to provide the results. RiskCardio produces a score which groups people according to the risk of cardiovascular death. The model deals with patients who have suffered from acute coronary syndrome (ACS). In simple words, the patients who survived a reduction or full blockage of coming blood to the heart.
Here is how RiskCardio works. First, the system divides ESG signals into pairs of consecutive beats. The researchers trained the system based on the data of past patients. They analyzed their ESG signals having divided the beats into groups. As a result, the pairs of adjacent beats of the patients who died were marked as “risky”, of the ones who didn’t die were marked as “normal”.
The main goal of the RiskCardio is to reduce the time needed for a doctor to make a decision. Obviously, heart diseases always require fast treatment and measures. AI could help doctors make faster decisions providing them with a score of a potential outcome. With the help of RiskCardio, doctors could get faster access to the history of patients and get a full picture of their health condition.
In addition, researchers are going to enhance the system. Such parameters as gender, age, lifestyle are crucial for choosing treatment measures. Therefore, if the system could analyze ECG signals taking into account age or overweight, it could provide more accurate predictions.