AI-driven algorithm used to detect atrial fibrillation much earlier

2,000 people took part in an innovative trial that tested an algorithm to help doctors identify patients at risk of atrial fibrillation.

The trial evaluates an algorithm named ‘FINDAF’ – which was developed by machine learning – that searches for red flags within GP records of people who may be at risk of developing AF over the next six-months.

The British Heart Foundation is funding the trial. It means that people who are identified using the algorithm will be offered additional testing to confirm a diagnosis.

The University of Leeds research team hopes that the West Yorkshire Pilot will set the foundation for a UK wide trial which could improve early diagnosis of AF one day and prevent more strokes.

In the UK, more than 1.6 millions people have been diagnosed as having AF. There are probably thousands of people living in the UK that have not been diagnosed and don’t know they suffer from the condition. Around 20,000 strokes occur in the UK each year as a result of AF.

The algorithm was integrated into the medical records of several GP practices in West Yorkshire. People who are identified as being at risk for AF will be offered home testing.

The ECG device is sent to those who consent. They are asked to perform two readings per day for four days, and at any other time when they experience palpitations. All of this can be done without the need to visit a GP.

The GP will be informed if the ECG results reveal that the patient has AF. They can then discuss possible treatment options.

BHF Clinical Director Dr Sonya Babu-Narayan said: “We offer effective treatments to people with atrial flutter who are at a high risk of stroke. Some people may not be aware that this hidden health threat is present.

This research, which uses routinely collected data on health care and predictive algorithms, offers an opportunity to identify people at risk of stroke and to offer them treatment.

Scientists and clinicians from the University of Leeds, Leeds Teaching Hospitals NHS Trust and the charity funded the development of the FIND-AF algorithm.

The team used the anonymous electronic health records from more than 2.1 millions people to train the algorithm. It was able to identify warning signs which indicate that the person is at high-risk of developing AF within the next six month.

The algorithm was validated by using data from more than 10 million people living in countries other than the UK. Leeds Hospital Charity is also funding the trial.