Artificial intelligence helps predict relapse of patients recovered from a heart attack or angina

  • This is the conclusion of a study published in The Lancet that has had the participation of IDIBELL and the Bellvitge University Hospital
  •  The algorithm, based on data from about 20,000 patients, will facilitate the choice of the best therapeutic option to avoid new episodes of heart failure
  •  The article establishes the basis for future studies in which patients can be treated according to the strategies recommended for these tools.
Unitat de Cures Intensives Cardiològiques de l'Hospital de Bellvitge [imatge d'arxiu]

New artificial intelligence tools will permit to predict the options that a person who has suffered a myocardial infarction or angina will die or suffer a new serious episode during the year after discharge. These are the conclusions of an international multicenter study that has just been published in The Lancet.

The study is led by specialists from the University of Turin (Italy) and, researchers from IDIBELL and the Bellvitge University Hospital also participated.

Patients who have recovered from acute coronary syndrome can have another heart attack, bleeding, or other complications. To prevent it, they often receive dual antiplatelet therapy, which reduces the risk of myocardial ischemia and increases the risk of bleeding. For this, it is very important to know the specific risk of each patient, in order to individualize the treatments.

The authors initially created and tested four new risk stratification models, obtained through data processing with machine learning systems. A total of 25 clinical, therapeutic, angiographic, and procedural variables extracted from 19,826 patients admitted between 2003 and 2016 in various centers around the world – among them the Bellvitge University Hospital – were introduced into these models, together with data on subsequent patients evolution.

The most effective model, called PRAISE, was then applied to another set of 3,444 patients and compared with the actual risk assessment. The final results have shown that the new tool predicts mortality, the risk of bleeding, and the risk of suffering a new heart attack much better.

Dr. Albert Ariza, one of the authors of the study, highlights that “the advantage of machine learning methods is that they apply algorithms to large and varied sets of data, capturing relationships between these data that simpler systems cannot capture.” Ariza, who is the coordinator of the Cardiology Intensive Care Unit of the Cardiology Service of the Bellvitge Hospital and researcher at IDIBELL, stresses that these are tools “are at initial steps, but they have already shown their usefulness in some other clinical; We hope that they will soon be available for all cardiologists and help us to make decisions based on personalized risk assessment”.

Now, the algorithm must be validated in future studies where patients receive the treatment recommended by PRAISE, especially regarding the duration of dual antiplatelet therapy.

Scroll to Top