Algorithm could help prevent thousands of strokes in UK each year
Share:
Tool uses machine learning to analyse medical data for red flags relating to undiagnosed atrial fibrillation. Thousands of strokes could be averted in the UK each year after doctors developed an algorithm that spots patients at risk by scanning millions of GP records.
The tool uses machine learning to scour medical data for red flags that could identify patients with undiagnosed atrial fibrillation (AF), a heart condition that significantly increases stroke risk. About 1.6 million people in the UK have been diagnosed with AF. Doctors believe there could be thousands more who are unaware they are living with the condition because they have no symptoms.
If diagnosed and treated early, the condition can be managed and the stroke risk reduced. The algorithm works out someone’s risk based on a number of factors including age, sex, ethnicity and whether or not they have other medical conditions such as heart failure, high blood pressure, diabetes, heart disease or chronic obstructive pulmonary disease (COPD).
Estimates suggest AF is a contributing factor in 20,000 strokes a year in the UK. The algorithm was developed by doctors and scientists at Leeds teaching hospitals NHS trust and the University of Leeds, with funding from the British Heart Foundation (BHF).
They developed the tool using the anonymised electronic health records of 2.1 million people, training the algorithm to find warning signs flagging people who are at risk of or may already have AF. The tool was validated with records from a further 10 million people.
