Israeli AI predicts heart failure in muscle inflammation patients with 80% accuracy
Myositis elevates chances of heart failure, so Rambam Hospital team built a special algorithm that analyzes ECG tests and gives early warning
By Nathan Jeffay December 2022, 4:55 pm
Illustrative: Doctors carry out heart bypass surgery after heart failure (KentWeakley via iStock by Getty Images)
Israeli researchers have built an artificial intelligence tool that analyzes ECG tests and predicts heart failure with high accuracy weeks before it happens.
The new technology is for sufferers of myositis — muscle inflammation — which elevates the risk of heart failure.
The physician who led the research, Dr. Shahar Shelly of Rambam Healthcare Campus, told The Times of Israel that it is the first AI tool built especially for this population. It analyzes heart patterns that are unique to them, and can bring about earlier detection than is currently possible, he said.
He reported in peer-reviewed research that the algorithm successfully predicted 80 percent of heart failure cases among a sample of myositis patients.
The study was carried out by running the algorithm on their ECG tests, and then comparing its predictions regarding who was at risk of heart failure with medical records to see who actually ended up suffering from heart failure.
Sign up for the Tech Israel Dailyand never miss Israel’s top tech stories Newsletter email address
By signing up, you agree to the terms
“We are running ECG tests through the AI model, which sees details that doctors can’t normally detect and then predicts who is at risk of heart failure,” said Shelly.
Dr. Shahar Shelly of Rambam Healthcare Campus (courtesy of Rambam Healthcare Campus)
“Given that it’s these cardiac dysfunctions that often end up killing people, this can save lives.”
Shelly conducted the study together with the researchers of the Cardiology Department at the prestigious US-based Mayo Clinic Medical Center.
- They taught the AI model by showing it ECG scans and medical records of 89 myositis patients from 2000 to 2020. The algorithm built a picture of subtle patterns in ECGs that seem to increase the chance of heart failure.
- It hasn’t been deployed in clinics yet, but Shelly said that this is the aim, after more research.
“Further down the road, the use of this model will allow the provision of appropriate treatments at an early stage, even before the deterioration of the patients’ medical condition,” he said. “We are talking here about preventing serious illness and even deaths.”
Shelly added: “Early detection is what is important here. The algorithm is looking for cardiac dysfunction, and when it finds it, doctors can take steps to prevent heart failure, such as changing treatment strategies.
“For a population that is at heightened risk of heart failure, this can make a very big difference.”