AI-Based Detection of Infantile Epileptic Spasms from Infant Movements in Videos
ENDEAVOUR PROJECT GRANT
£174,981.70
Dr Edmond Shu Lim Ho
University of Glasgow
Infantile Epileptic Spasms (IES) are a serious and time-sensitive form of epilepsy that occurs in babies. Early and accurate diagnosis is essential for effective treatment and better long-term outcomes. However, recognising IES is difficult, even for specialists, especially when relying on brief observations or low-quality video recordings. While a few studies have used artificial intelligence (AI) to detect seizures from video footage, these were conducted in controlled hospital settings with very small numbers of babies. Therefore, there is a need for a reliable, real-world tool that can support diagnosis using videos recorded at home by parents or carers.
"This study could be transformative for people affected by epilepsy, particularly infants and their families. The AI system could significantly reduce the time to diagnosis, allowing for earlier intervention and improved treatment outcomes. Additionally, the remote, video-based nature of the system can provide access to specialized assessments for those in underserved or remote areas, reducing the burden on healthcare systems and improving the quality of life for affected individuals and their families.
This project will develop an AI-based system to detect signs of Infantile Epileptic Spasms from videos recorded on smartphones by families in everyday settings. The research will use the NHS-approved vCreate Health Web-App to securely collect video recordings from caregivers. These videos will then be assessed and labelled by clinical experts to identify key body movement patterns linked to IES.
Using this labelled data, researchers will train an AI tool to automatically recognise signs of IES. This will be based on a secure cloud computing environment to ensure the safety and privacy of all patient data. Patient and public involvement (PPI) activities will be central to the project, helping to co-design a user-friendly tool that supports caregivers in capturing high-quality, clinically useful videos.
The end goal is to integrate this AI system into a web-based decision support tool to aid, not replace, clinicians in diagnosing IES more effectively.
This research has the potential to transform the way IES are detected. By enabling accurate analysis of everyday smartphone videos, it could shorten the time to diagnosis and improve access to expert evaluation, especially for families living in remote or underserved areas.
Importantly, the tool is designed to work alongside healthcare professionals, offering support rather than replacing clinical judgement. If successful, this system could reduce pressure on healthcare services, enhance early intervention, and improve the quality of life for infants with epilepsy and their families.
