Kantify and the ULB announce a common breakthrough in the prediction of Atrial Fibrillation
Atrial Fibrillation, a cardiac disease affecting 2% of the world population, can now be predicted.
Atrial Fibrillation, a pathology that affects 2% of the total world population, is largely un- or underdiagnosed.
Over the past year our team has worked in close collaboration with Dr Jean-Marie Gregoire, cardiologist and IRIDIA, the AI Laboratory of the Université Libre de Bruxelles. Together, we have successfully developed an AI-based model that can predict an oncoming event of Atrial Fibrillation in patients. This is particularly noteworthy, because it was not previously known that Atrial Fibrillation events showed any symptoms before their occurrence.
The models developed by Kantify and the ULB make it possible to predict Atrial Fibrillation at individual level, without any history or data about the patient, except for their RR-intervals collected through a holter monitor. The models are trained on a dataset of more than ten thousand individual anonymized cardiac monitorings of a duration between 24h and 36h, collected and annotated over multiple years. The balanced accuracy of the model is 80%, 30 seconds before the atrial fibrillation event occurs.
In the media :