New AI tool “can quickly rule out heart attacks in people visiting the emergency room”

A new artificial intelligence (AI) tool developed in the UK can quickly rule out heart attacks in patients attending the emergency room, saving tens of thousands of people from unnecessary hospital stays each year, according to its developers.

The AI ​​tool, known as Rapid-RO, was able to successfully rule out heart attacks in over a third of patients during test runs in four British hospitals.

Professor James Leiper, deputy medical director of the British Heart Foundation (BHF), which funded the study, said: “This research shows the important role that AI could play in treatment decisions for heart patients.

“By quickly identifying patients who can be safely discharged, this technology could help prevent unnecessary hospital admissions, allowing valuable NHS time and resources to be spent where they can be of greatest benefit.”

When a patient is admitted to hospital with suspected heart attack, the diagnosis is generally confirmed by blood tests.

These tests measure the level of a protein called troponin, which increases when there is damage to the heart muscle.

However, because this increase may not be reliably detectable until hours later, patients often need to remain in the hospital for further troponin testing and monitoring.

Some of these patients are eventually discharged without requiring treatment after a heart attack is ruled out.

Dario Sesia, a BHF-funded PhD student at Imperial College London, developed Rapid-RO to identify patients who are at very low risk of heart attack.

Rapid-RO was trained using data from over 60,000 patients across the UK and subsequently tested on more than 35,000 patients.

The data from the first troponin blood test is combined with other patient information collected during the hospital stay and then analyzed by the algorithm.

The patients are then assigned either a very low risk of a heart attack or no risk at all.

Rapid RO was able to successfully rule out heart attacks in 36% of patients, compared to 27% of patients using a troponin blood test alone.

In addition, it has been found that heart attacks can be detected more accurately.

Troponin testing missed four times as many heart attacks (108 cases) as the AI ​​tool (27 cases), the researchers said.

In addition, the method was effective regardless of ethnicity and gender, as well as whether the patients had Covid-19, they added.

Dr Amit Kaura, Postgraduate Clinical Research Fellow in Cardiology at Imperial College London, said: “Current methods for ruling out a heart attack combine a clinical examination with a blood test to measure troponin, a blood marker of heart muscle damage.

“Many patients require multiple troponin tests to confirm they have not had a heart attack, resulting in longer hospital stays and higher costs.

“We have developed an artificial intelligence-based model that uses age, initial blood tests – including troponin – and other basic health information to help doctors rule out heart attacks more quickly than current methods, while maintaining high accuracy across different age groups and in patients with different health conditions.

“Our study shows how artificial intelligence can help physicians make more timely patient care decisions, avoiding unnecessary hospitalizations while ensuring patient safety.”

In the next step, the researchers want to convert Rapid-RO into an app that can be used by doctors.

Professor Leiper said: “We look forward to further research to understand how Rapid-RO could be used in the future to accelerate clinical decisions and improve patient treatment and care.”

The results were presented at the British Cardiovascular Society conference in Manchester.