Artificial intelligence can detect heart-disease risk in less than two seconds. Doctors have developed a technology that analyzes results from a standard ECG test to extract information invisible to the human eye.
The technology can detect signs of heart failure and heart valve disease. This could allow at-risk patients to be identified quickly and prioritized for echocardiograms.
Details of the AI technology, trained on data from several million ECGs, were presented at the European Society of Cardiology’s annual congress. Thousands of doctors and researchers from around the world attended the conference, held in Munich, Germany. It is the world’s largest cardiology conference.
ECGs will now reveal extra information
An electrocardiogram, or ECG, records the heart’s electrical activity. It provides information on the heart’s rate and rhythm. The test has been used for nearly a century to detect heart attacks and abnormal heartbeats.
However, a standard ECG cannot definitively detect heart failure or heart valve disease. Diagnosing these conditions generally requires an echocardiogram, or an ultrasound of the heart.
Patients often have to wait several months for this test.
The new AI technology could help shorten that wait. By analyzing ECG results, it can quickly identify possible signs of these conditions.
Tested on 67,000 patients
The technology was tested on 67,000 patients in the United States.
The study found the AI could detect up to 81 percent of patients with heart failure. For heart valve disease, the detection rate was as high as 90 percent.
However, the AI alone cannot make a final diagnosis of any condition. Nor can it definitively say someone does not have these conditions.
Instead, it can give a strong indication of whether someone is more likely to have the condition.
If the risk appears high, that patient could be sent quickly for an echocardiogram. This could allow patients to skip long waiting lists and get tested quickly.
As a result, diagnosis would be faster, and necessary treatment could begin earlier.
‘Early detection saves lives’
The research was funded by the British Heart Foundation (BHF). The foundation’s consultant cardiologist and clinical director, Sonya Babu-Narayan, said it was “exciting” that AI could deliver results this quickly from an ECG.
She said the technology would not be able to detect everyone with heart disease. However, she said, it could help quickly identify high-risk patients and prioritize them for necessary tests.
She said that if heart disease is detected quickly and treatment begins promptly, it is possible to save lives and improve quality of life.
Nearly 1 billion ECGs a year
A major reason for interest in the technology’s potential is the widespread use of ECGs.
Roughly 1 billion ECG tests are performed worldwide every year. This means that by using data from this routine test performed at hospitals and medical centers, it could become possible to identify heart-disease risk in large numbers of people in advance.
Imperial College London cardiology professor Fu Siong Ng said patients referred for an echocardiogram on a doctor’s advice often have to wait several months.
He said that if the new technology can identify at-risk patients, they could be sent for scans more quickly and on a priority basis.
Heart disease could also be caught during other tests
Professor Ng also spoke about another potential use for the AI.
Often, a patient’s ECG is taken for other reasons even when there is no suspicion of heart disease. That ECG data could also be analyzed using AI.
If every ECG performed at a hospital could be analyzed through the AI model, it would become possible to separately identify high-risk patients.
This could also allow previously unsuspected heart failure and valve disease to be detected quickly.
A handheld AI ECG device
The analysis was led by Imperial College London researcher and BHF clinical research fellow Ahmed El-Medany.
He described the technology as “superhuman AI.”
Researchers’ current goal is to develop a handheld, AI-powered ECG device that health workers can use.
If such a device is developed, it could become possible to quickly detect heart-disease risk outside hospitals or at primary care centers.
High blood pressure, diabetes from a 5-second facial video
Details of another potential medical use for AI were also presented at the Munich conference.
Researchers from the University of Tokyo and the Institute of Science Tokyo said that analyzing just a five-second video of a person’s face could make it possible to detect high blood pressure and type 2 diabetes.
According to the researchers, this AI-based analysis could make diagnosis easier.
Worldwide, millions of people with high blood pressure and type 2 diabetes are unaware they have the conditions.
Researchers hope that, in the future, this kind of technology could help quickly identify such undiagnosed patients and bring them under necessary treatment.