Development of a new artificial intelligence that predicts the death of heart patients

Researchers have developed a new artificial intelligence tool that can help doctors predict the conditions of high-risk heart patients; This artificial intelligence can predict the risk of death by examining the patient’s heart scan.

According to a Euronews report, researchers at Imperial College London developed the artificial intelligence model AIRE, which was trained with millions of electrocardiogram (ECG) results. An electrocardiogram is a common medical test that records electrical signals in and between the chambers of the heart and is usually done to diagnose heart attacks and other heart irregularities.

New artificial intelligence to predict the death of heart patients

The goal of developing this artificial intelligence is to identify subtle patterns in a patient’s tests that may mean someone is at high risk or dying. For testing, this model was able to predict the probability of death a decade earlier according to the patients’ ECG records, and it was correct in 78% of the cases.

Of course, according to the researchers, predicting the patient’s risk of death with the help of artificial intelligence may turn out to be wrong; Because the patient may undergo certain treatments or die unexpectedly earlier than expected, this model can detect subtle changes in the structure of the heart, which may be a warning to the patient, but this sign may be hidden from the eyes of doctors.

The new system can also predict heart attacks, heart failure and heart rhythm problems. Researchers say the system could be rolled out across the UK’s National Health Service (NHS) within the next five years. Trials on real patients are expected to begin by mid-2025.

Of course, according to the researchers, more research is needed to complete this model for more accurate diagnosis and treatment; Also, patients who have other diseases can probably benefit from this model; For example, diabetes also affects the heart.

The researchers’ research findings have been published in Lancet Digital Health.