Artificial intelligence predicts if and when someone will go into cardiac arrest

A new method based on artificial intelligence can predict, far more accurately than a doctor, whether a patient might die of a heart attack and when, according to the researchers published in the journal ‘Nature Cardiovascular Research’.

Built by scientists at Johns Hopkins University (United States) from raw images of diseased hearts and patient history, technology can revolutionize clinical decision-making and increase survival from sudden and fatal cardiac arrhythmiasone of the deadliest and most baffling conditions in medicine.

“Sudden cardiac death caused by an arrhythmia represents up to 20% of all deaths in the world and we know little about why it occurs or how to know who is at risk,” explains lead author Natalia Trayanova, professor of Biomedical Engineering and Medicine Murray B. Sachs.

“There are patients who may be at low risk of sudden cardiac death and who get defibrillators that they may not need, and then there are high-risk patients who don’t get the treatment they need and could die in the prime of life,” he adds. . What our algorithm can do is determine who is at risk of cardiac death and when it will occurwhich allows doctors to decide exactly what needs to be done.”

The team is the first to use neural networks to build a personalized survival assessment for each heart disease patient. These risk measures provide with great precision the probability of a sudden cardiac death over 10 yearsand when it is most likely to occur.

Deep learning technology is called Survival Study of Cardiac Arrhythmia Risk (SSCAR). The name alludes to cardiac scarring caused by heart disease that often results in fatal arrhythmias, and the key to the algorithm’s predictions.

The team used Contrast-enhanced cardiac images visualizing the scar distribution of hundreds of real Johns Hopkins Hospital patients with cardiac scars to train an algorithm to detect patterns and relationships not visible to the naked eye. Current clinical cardiac image analysis only extracts simple scar features, such as volume and mass, underutilizing what has been shown in this paper to be critical data.

“The images contain critical information that clinicians haven’t been able to access,” says first author Dan Popescu, a former Johns Hopkins doctoral student. This scarring can be distributed in different ways and says something about a patient’s chances of survival. There is hidden information in it.”

The team trained a second neural network to learn from 10 years of standard clinical patient data, 22 factors such as patients’ age, weight, race, and prescription drug use.

Not only were the algorithms’ predictions significantly more accurate on every measure than the doctors’, but were validated in tests with an independent cohort of patients from 60 health centers across the United States, with different cardiac histories and different imaging datasuggesting that the platform could be adopted anywhere.

“This has the potential to significantly shape clinical decision-making in relation to arrhythmia risk and represents an essential step in bringing patient trajectory prognosis into the era of artificial intelligence. –says Trayanova, co-director of the Alliance for Innovation in Cardiovascular Diagnosis and Treatment–. It is the epitome of the trend to merge artificial intelligence, engineering and medicine as the future of healthcare.

The team is now working on the creation of algorithms to detect other heart diseases. According to Trayanova, the concept of deep learning could be developed for other fields of medicine that rely on visual diagnosis.