‘Superhuman’ AI that predicts risk of premature death to be tested in UK

LONDON: A “superhuman” artificial intelligence model designed to predict a patient’s risk of illness and premature death will be trialled in UK hospitals next year.

The technology, which uses readings from a common, cheap heart test to alert doctors to patients who may benefit from further testing or treatment, is expected to be used across the health service within five years.

The AI ​​program, known as AI-ECG risk estimation, or Aire, was developed to read the results of electrocardiogram (ECG) tests, which record the electrical activity of the heart and are administered to patients suspected of having heart problems.

It then uses these recordings to detect problems in the structure of the heart that doctors would not be able to see.

The technology will be trialled for the UK’s National Health Service (NHS) at Imperial College Healthcare NHS Trust and Chelsea and Westminster Hospital NHS Foundation Trust from mid-2025, with further hospitals to be confirmed.

Dr Fu Siong Ng, professor of cardiac electrophysiology at Imperial College London and consultant cardiologist at Imperial College Healthcare NHS Trust, told the PA news agency: “There are three or four different studies that will be done in hospitals starting in middle of next year. year.

“Generally speaking, they must prove that these models are accurate in detecting diagnoses. So, for people who already had an ECG in the hospital, we would prove that these models are accurate in detecting certain diagnoses.

“Then in the next layer, once that’s been demonstrated, we can intervene to show that we can change the trajectory of patients.”

It is understood that several hundred patients will be recruited for the first trial and that the number will then be expanded for subsequent studies.

Dr Ng added: “The vision is that every ECG performed in the hospital will be subjected to the model.

“So anyone who has an ECG anywhere in the NHS in 10 years, or five years from now, will be subjected to the models and doctors will be informed, not only about what the diagnosis is, but about a prediction of a set complete”. range of health risks, meaning we can intervene early and prevent disease.

“If, for example, it says you are at high risk for a specific heart rhythm problem, you might be more aggressive in preventive treatment to prevent this from happening.

“There are some related to weight, so you can subject them to weight loss programs.

“You could even think about earlier medical treatments to prevent things from progressing, but that will be the subject of the clinical studies we plan to do.”

Dr Arunashis Sau, a British Heart Foundation (BHF) clinical researcher at the National Heart and Lung Institute at Imperial College London and cardiology registrar at Imperial College Healthcare NHS Trust, told PA that Aire’s aim It is not to develop something to replace doctors, but to create something “superhuman.”

“The goal here is to try to use the ECG as a way to identify people who are most at risk, who will then perhaps benefit from other tests that could tell us more about what’s happening,” he said.

“The ECG is a very common and very cheap test, but it could be used to guide more detailed tests that could then change the way we manage patients and potentially reduce the risk of something bad happening.

“A key distinction is that the goal here was to do something that was superhuman, that is, not to replace or speed up something that a doctor could do, but to do something that a doctor cannot do based on heart monitoring.”

It comes after research published in Lancet Digital Health found that Aire was able to correctly identify a patient’s risk of death in the 10 years after the ECG, from highest to lowest, in 78% of cases.

For the study, the team trained Aire using a data set of 1.16 million ECG test results from 189,539 patients.

The platform could also predict future heart failure in 79% of cases, future serious heart rhythm problems in 76% of cases, and future atherosclerotic cardiovascular disease (where the arteries narrow, making blood flow difficult) in 70%. of the cases.

Dr. Sau said: “We know that not only doctors, but also patients must trust AI. And that’s a big part of the work we did here.

“What we found is that the AI ​​could pick up things related to the structure and function of the patient’s heart, and even things as profound as genetic information were collected that the AI ​​could use in combination to find out that these people might be at higher risk, and these are things that are not obvious to human doctors.” – dpa/Tribune News Service