Scientists have created an algorithm using artificial intelligence to predict the risk of heart attack or stroke.
According to the Radiological Society of North America, Deep Learning algorithms are trained on large databases and are able to notice the smallest connections between image details and certain facts.
The model offers a method for population-based screening of cardiovascular disease risk using conventional X-rays, says Jacob Weiss, one of the authors of the development. It can be used to identify people who statistically need treatment.
The current version of the algorithm estimates the risk of disease over a 10-year period, and can determine which patients need to take statins - drugs to lower cholesterol. A total of 147,497 chest X-rays from 40,643 patients were used for training. After training was completed, the authors tested the algorithm on archived medical data from 11,430 patients.
1,096 of them (9.6%) had experienced serious cardiovascular problems like strokes and heart attacks within the next 10 years of follow-up. It turned out that the algorithm would have been capable of predicting the occurrence of these problems with high accuracy, meeting accepted standards of diagnosis.