A new artificial intelligence tool, known as FIND-AF, has emerged with the potential to reshape the landscape of heart condition detection, specifically targeting atrial fibrillation (AF). Developed by a collaborative team of scientists and clinicians at the University of Leeds and Leeds Teaching Hospitals NHS Trust, this innovative system utilises extensive anonymised health records to identify individuals at risk of developing AF even before symptoms manifest.
Atrial fibrillation is characterised by an irregular and often rapid heartbeat, and can significantly elevate the risk of stroke. Although there are currently around 1.6 million diagnosed cases in the UK, experts from the British Heart Foundation (BHF) suspect that many individuals remain undiagnosed, unaware of their vulnerability to stroke.
FIND-AF employs machine learning algorithms to analyse medical records, assessing various risk factors such as age, sex, ethnicity, and existing health conditions like heart failure, hypertension, diabetes, ischaemic heart disease, and chronic obstructive pulmonary disease. In total, the algorithm has been trained on 2.1 million anonymised health records and validated against an additional 10 million records.
Once identified as being at high risk, patients are equipped with a handheld electrocardiogram (ECG) device designed for home use. Over a four-week period, they are required to take readings twice daily, or any time they experience symptoms such as palpitations. Results are sent directly to the trial team for analysis. If the ECG readings indicate irregular heart rhythms consistent with AF, a notification is dispatched to the patient’s general practitioner for further discussion regarding treatment options. These may involve medications to mitigate stroke risk or lifestyle modifications aimed at improving heart health.
One participant, 74-year-old John Pengelly from Apperley Bridge, West Yorkshire, was diagnosed with AF after participating in the trial, despite exhibiting no obvious symptoms. "I got a letter inviting me to take part in the study and thought: 'If it benefits somebody, then great, I want to help,'" he recounted. Following his involvement, Pengelly began taking daily medication to reduce his stroke risk, expressing gratitude for the proactive measures that identified his condition.
Atrial fibrillation occurs when the upper chambers of the heart contract erratically, hampering the organ's ability to relax properly between beats and thus diminishing its efficiency. The precise causes of AF remain unclear, but it is more prevalent in older adults and those with chronic conditions like heart disease or obesity. Potential triggers include excessive alcohol consumption or smoking.
Chris Gale, a professor of cardiovascular medicine at the University of Leeds, remarked, "All too often the first sign that someone is living with undiagnosed atrial fibrillation is a stroke." He emphasised the devastating impact this can have, not just on patients but also on healthcare systems, which could incur significant costs by failing to identify and treat AF expediently.
Dr Sonya Babu-Narayan, associate medical director at the British Heart Foundation, echoed these sentiments, highlighting that "effective treatments are available for individuals at high risk of stroke due to AF." She underscored the potential of using routinely collected healthcare data and predictive algorithms to uncover more cases of this "hidden threat" to health.
Dr Ramesh Nadarajah from the Leeds Teaching Hospitals NHS Trust further noted the vast potential of data collected during patient interactions with the NHS to facilitate earlier identification and testing for conditions like AF. He expressed optimism that the study could lead to a larger nationwide trial aimed at incorporating this algorithm into standard clinical practice.
The prospects of integrating AI into healthcare gained notable attention from Health Secretary Wes Streeting, who recently described the use of AI, machine learning, and big data as "game-changing" for improving healthcare outcomes. He posited that such technologies could allow for earlier diagnosis and treatment, ultimately revolutionising the prevention of illness.
With the advancement of tools like FIND-AF, the goal remains clear: to identify individuals at risk of atrial fibrillation at an earlier stage and enhance treatment options, thereby potentially reducing the substantial likelihood of stroke attributable to undiagnosed heart conditions.
Source: Noah Wire Services