How AI can predict heart attacks ten years in advance – 11/24/2023 – Science

Cardiovascular diseases, which continue to be the main cause of death in the world, claim more than 17 million victims every year. And it is estimated that this number could reach 23.6 million by 2030.

Although the number of cardiovascular diseases continues to increase, the death rate from heart attacks has decreased significantly, thanks to preventative measures and better emergency medical care.

With the help of artificial intelligence (AI), this exciting trend is expected to gain even more momentum in the future. This is because AI is able to detect potential threats in time, and this is its great asset: from a variety of exam data, AI can find specific patterns or irregularities much faster than the most well-trained professionals.

In the future, AI could save the lives of thousands of patients with chest pain and whose risk of a heart attack has not been detected on a CT scan – often, subtle narrowings in the coronary arteries end up being missed on exams, but, in In case of inflammation, they can rupture and block the arteries, causing a heart attack.

Until recently, it was not possible to identify these high-risk patients in advance. Now, an AI tool developed by Oxford researchers can detect even the most subtle abnormalities in coronary arteries and predict the risk of heart attack, over the next ten years, in a patient with chest pain. The result of a study funded by the British Heart Foundation, the invention was presented at a scientific panel at the American Heart Association in Philadelphia, United States.

Many high-risk patients go unnoticed

For the study, the team led by Professor Charalambos Antoniades, from the Radcliffe Department of Medicine at the University of Oxford, analyzed data from more than 40,000 patients from eight British hospitals who had a heart scan as part of a routine exam.

Cardiac tomography is a standard exam to detect narrowings or obstructions in the arteries that supply the organ. In the United Kingdom alone, approximately 350,000 people per year take this exam.

About 3 out of 4 cases analyzed did not show clear signs of significant narrowing in the coronary arteries, and the patients were discharged without further treatment. But analysis of the data by Antoniades and his team revealed that a portion of these released people, contrary to what was believed, were especially vulnerable: the incidence of heart attack or heart failure followed by death in this group was doubled.

These patients were not recognized as part of the risk group because small signs went unnoticed. AI has made visible what man could not see: the tool developed by Oxford researchers calculates the real risk of heart attack by examining changes in the fat around inflamed arteries, as it is these changes that point to a significantly higher risk of heart attack .

More targeted and efficient treatment

In a world-first pilot project, Oxford scientists developed AI for risk assessment in 744 patients. Based on this, doctors changed the treatment in 45% of cases. The implementation of the technology, the study authors say, reduced the number of heart attacks by more than 20% and led to 8% fewer deaths and strokes.

“Each year, many people die unnecessarily from a heart attack. It is very important that we take advantage of the potential of artificial intelligence to guide patient treatment,” says Antoniades. “Here, we show that providing clinicians with an accurate view of risks can alter and possibly improve the course of treatment.”

Saving time and money

The newly developed AI tool is just one example of how artificial intelligence can help with early treatment – ​​something that can save lives, save a lot of time and also a lot of money.

Another example is an AI-based “decision support system” developed by British cardiologists in mid-2023, something that could make it easier to diagnose a heart attack in the emergency room.

The tool is capable of determining the value of troponin in the blood – a protein released by decomposing cardiac muscle cells and which plays a central role in the diagnosis of heart attack – much faster than conventional methods. Until now, patients are often taken to the hospital or have to stay there for longer observation, even if they have not suffered a heart attack, until the troponin value is determined. In cases like this, AI could save time and unnecessary expenses.

Early identification of risks

AI not only detects risks faster, but in some cases does so more reliably than humans.

Also in mid-2023, American researchers presented artificial intelligence that looks for suspicious patterns in the electrocardiogram curve and is capable of detecting dangerous obstructions in the coronary arteries more reliably than even experienced cardiologists.

If the danger is detected earlier thanks to AI, the risks decrease. Long before a dangerous obstruction occurs, it is possible, through a relatively safe routine procedure, to implant a stent in the patient – ​​a small metal or plastic pipe that, when taken to the narrow point of the blood vessel, keeps it permanently open, preventing narrowing and, consequently, reducing the risk of a heart attack.