Artificial intelligence predicts the possibility of a new heart attack or bleeding in heart attack patients.
A revolutionary approach, the result of a research carried out by the University Cardiology of the Molinette hospital of the City of Health of Turin directed by Gaetano Maria De Ferrari, in collaboration with the University and Polytechnic of the Piedmontese capital, published today in the Lancet magazine.
"The study, thanks to which we can now better treat our patients, is a very strong demonstration of the possibilities of Artificial Intelligence in medicine", explains De Ferrari.
"Patients with acute myocardial infarction - explains Fabrizio D'Ascenzo, coordinator of the study - are at very high risk in the first two years of both a recurrence of heart attack and major bleeding related to drugs that keep the blood 'more fluid', such as Cardioaspirin. The decision on the best therapy must balance these two risks, which the cardiologist does based on his own experience and clinical intuition, aided by risk scores. However, these scores are imprecise and therefore of modest help even for a cardiologist. We therefore tried to improve the situation by using clinical data on 23,000 patients, many of them collected in Piedmont, which provided the critical mass of information for our research. "
While the accuracy of the best available scores to identify the possibility of an event such as a new heart attack or bleeding hovers around 70%, the accuracy of this new risk score approaches 90%, statistically reducing the chance of an incorrect diagnosis from three to one patient out of ten analyzed.
(HANDLE).