A new artificial intelligence (AI) algorithm can predict, up to 13 minutes in advance, the possibility that a hospitalized patient will suffer an acute myocardial infarction, allowing doctors to act before the infarction occurs.
The system was developed by Iscte – Instituto Universitário de Lisboa, which created the AI models, and by the Faculty of Medicine of the University of Lisbon (FMUL), which provided the clinical data that trained and tested these models.
Tested on patients at the Santa María Hospital in real cardiology intensive care conditionsThe algorithm allows a “Significant anticipation” of the possibility of an acute myocardial infarctionwhen analyzing the risk of ischemia (decreased flow of oxygen-rich blood to the heart muscle), Miguel Sales Dias, principal investigator of the AIM Health project, told Lusa.
According to the Iscte researcher, this new AI algorithm was developed and tested together with the cardiologist and FMUL professor, Luís Rosário, and his team from the cardiology intensive care unit of the Santa María Hospital, with the data being submitted to the ethics committees. of both institutions.
In practice, a series of physiological data are collected and monitored from a patient and the algorithm is invoked permanently. As soon as it detects a relevant change in physiological signs, it raises the alarm to the responsible doctor, explained Miguel Sales Días.
“We already developed a pilot, which had good results, and now we are working to put it into operation,” said the Iscte researcher, adding that the intention is to expand the system to other cardiology intensive care units.
With the use of artificial intelligence in this area, “we have already entered the field of prediction well in advance,” he highlighted.
“Until now, patients are monitored and, when things happen, measures are taken,” cardiologist Luís Rosário told Lusa, for whom the new AI system could lead, in this area, to move from reactive to proactive action. . action.
This means that “Before the patient has a heart attack, before the patient’s heart suffers and is damaged, we can start therapy or correct the damage that causes it.“said the FMUL professor.
The expert also stated that the results, which will be presented on Wednesday at the “New Frontiers in Health” conference dedicated to the challenges of public policies in the digital age, in Lisbon, are “very promising”, but now it is necessary to improve the system reliability to have clinical impact.
In addition to the ability to predict impending ischemia, the AIM Health project has developed a second advanced tool: a computational system that uses computer vision and artificial intelligence technologies to help cardiologists diagnose aortic valve calcification problems.
This system automatically identifies the valve in MRI and echocardiography examinations, having the ability to distinguish patients with a calcified valve from healthy patients and also quantify the calcium present.
“The set of images that we use to train our algorithm, if the doctor were to observe them manually it would take 22 hours and therefore it is real time“, highlighted Miguel Sales Días.
For Luís Rosário, the speed and precision of this technology facilitate early diagnosis, essential for planning personalized treatments, but also allow doctors to save time and focus on more complex clinical tasks.
Source: Observadora
