The partners will develop a deep learning system that can detect a stroke immediately after facial paralysis occurs in a patient. Facial paralysis is a loss of facial movement due to nerve damage, and around 40 % of those experiencing a stroke will also experience this symptom. When this occurs it is important that effective treatment is initiated as soon as possible to reduce the risk of fatal consequences. At this moment, there are no pre-hospital aids that can detect a stroke when it occurs, or in the early stages. 

In this project titled “Instant recognition of stroke using artificial intelligence”, there will be developed an algorithm using image analysis that can identify a stroke with facial paralysis almost immediately after it occurs. The project is led by Gisle Halvorsen at LHL hjerneslag and will begin on the 1st of March. They will receive a funding of 2 million NOK from stiftelsen DAM. Michael Riegler from the Department of Holistic Systems at SimulaMet will assist Halvorsen in the development of the system with his experience and scientific knowledge in the intersection of artificial intelligence and medicine.

– I will support the project with advice based on my experience working with AI and healthcare for several years. This will range from data collection, over model development and evaluation to interpretation, presentation and explanation of results, says Riegler. 

The aim is to shorten the time it takes from when a facial paralysis occurs to when the patient receives treatment. In addition, they aim to make the patient immediately aware of the severity of this symptom. 

Professor Martin Kurz leads the stroke unit at Stavanger University Hospital and will be responsible for overlooking and facilitating the necessary expertise from the hospital. 

– The project is innovative and groundbreaking. The project has the potential to develop a tool that enables ultra-early detection of stroke symptoms using artificial intelligence. Such a tool can have a huge impact on future treatment methods, says Kurz. 

The project is funded by Stiftelsen DAM. 

The partners are LHL hjerneslag, ResQ Biometrics and Stavanger University Hospital (SuS).