Patients at risk of serious heart disease could soon be identified in less than two seconds using a routine heart test and artificial intelligence. The technology may help doctors decide who urgently needs further checks, instead of making every patient wait months for a scan.
The AI tool analyses electrocardiograms, commonly known as ECGs. These tests record the heart's electrical activity, including its speed and rhythm. Doctors have used ECGs for about a century to diagnose problems such as heart attacks and unusual heart rhythms. However, an ECG alone cannot normally detect heart failure or heart valve disease.
Those conditions are usually found with an echocardiogram, an ultrasound scan that produces images of the heart. In some health systems, patients may wait several months for this scan. An earlier diagnosis can be vital because it allows doctors to begin treatment before a patient becomes dangerously ill.
Researchers trained the new system using information from millions of patients. It can find patterns in an ECG that are difficult for the human eye to see. In a US trial involving 67,000 patients, the tool identified up to 81% of people with heart failure and up to 90% of those with heart valve disease.
The technology is not designed to make a final diagnosis by itself. It also cannot prove that a person has no heart condition. Instead, it gives doctors a strong indication of which patients are most likely to have a problem. Those at higher risk could then be moved ahead in the queue for an echocardiogram.
Professor Fu Siong Ng of Imperial College London said the tool could help hospitals prioritize patients who need scans more quickly. It might also discover hidden disease in people who received an ECG for another reason. Hospitals could run the AI analysis on all ECG results and flag patients who appear to be at high risk, even when doctors had not suspected heart disease.
The British Heart Foundation, which funded the trial, said the system would not find every person with a heart condition. Still, it could help doctors identify high-risk patients sooner. With about one billion ECGs carried out worldwide each year, adding AI analysis to this common test could have a wide impact.
The findings were presented at the European Society of Cardiology's annual congress in Munich. The researchers' next challenge is to develop handheld ECG readers with built-in AI that healthcare professionals could use more widely.