Study shows routine use of Artificial Intelligence-based machine learning algorithms could help dramatically increase detection rates of people with undiagnosed atrial fibrillation

Uxbridge, Middlesex & Walton Oaks, Surrey November 4, 2019 – An artificial intelligence (AI)- based machine learning (ML) technique has been shown in a test database to exhibit greater predictive performance than other currently available risk-prediction models for atrial fibrillation (AF) such as CHARGE-AF, according to the findings of a UK study published in PLOS ONE.ⁱ

The study found that the algorithms, which were developed using routine patient records, have the potential to enrich the patient population for targeted screening.ⁱ The next stage of research is to validate how the algorithms perform in clinical practice with the hope being that it can reduce the number of patients needed to be screened.

In the UK, up to 300,000 people are estimated to be living with undiagnosed AFⁱⁱ, which increases their risk of a stroke fivefold.ⁱⁱ The condition is underdiagnosed because of a lack of cost-effective, routine screening methods, and because people with AF may only experience sporadic symptoms, or none at all.ⁱⁱ

Current methods for AF detection, such as opportunistic pulse checking in the over 65 years and over age group, mean that around 100 people are screened in order to identify one person living with AF.ⁱⁱⁱ The study found that adopting the AI algorithm could reduce this number to 1 in 9.ⁱ The study tested whether AI was more accurate than existing risk prediction models using the health records of nearly 3 million people.

“This AI technique represents quite an astonishing leap in precision,” said Professor Mark O’Neill, one of the authors and Consultant Cardiologist at St Thomas’ Hospital and King’s College, London. “The implications are huge, especially because ML can be so easily and affordably used in routine clinical practice with the potential to transform the diagnosis of AF. If we can find and treat people living unwittingly with AF, we can do a much better job of preventing complications like stroke and heart disease.” 

The ML algorithm is potentially more precise than routine practices because it not only looks for risk factors, but also how they change, and can spot complex relationships between ‘risk predictors’, that cannot be readily identified by humans, such as subtle changes in blood pressure prior to diagnosis or frequency of GP visits.ⁱ

The next step is to test the algorithm in routine clinical practice and quantify its impact in terms of the number of AF cases identified and the associated potential cost savings in the earlier detection of AF.