Last month’s SophI.A Summit brought together the world’s experts in Artificial Intelligence (AI) and health.
Academics, doctors and leaders from across the business world discussed AI research and emerging applications in the life sciences sector.
Professor Sébastien Ourselin of King’s College London explained how a consortium of London universities is building an AI and big-data infrastructure across 10 hospital trusts to deliver on a vision for AI-enabled care.
The partner hospitals provide care to a quarter of the UK’s population, so the London AI Centre will consolidate learning from up to 16 million patients.
The infrastructure, built with privacy in mind, will enable all clinical data (i.e. imaging, genetics, blood, pharmacy, interventions, medical decisions and costs) to be used for research, clinical and operational purposes, allowing algorithms to be written without data ever leaving the hospital.
Professor Ourselin said: “To achieve this, we are building a large computing infrastructure within each hospital that allows the integration of AI models into clinical workflows, translating AI technologies into the real world A centralised orchestration system will enable federated learning across multiple hospitals.”
He went on to introduce a large-scale research programme in Acute Neurology, built and deployed around this platform. This demonstrates how to make sense and extract value from real-world neurological and neuroradiological clinical data using advanced AI models, to enable the creation of new clinical tools that inform, augment, and improve the treatment of patients with acute neurological conditions.
Juliette Raffort, consultant in medical biology in the Department of Clinical Biochemistry at CHU Nice led a session on AI applications in vascular diseases.
Cardiovascular diseases are the leading cause of premature death in developed countries and abdominal aortic aneurysm (AAA) has become a significant public health challenge worldwide. A key problem is a critical lack of tools to evaluate patient outcomes.
Raffort explained the development of AI-derived applications for patients with vascular diseases, mainly focusing on aortic aneurysm. While AI in non-cardiac vascular diseases is still in its infancy, her team is using innovative techniques in biology and imaging to identify new patterns, combined with AI to develop efficient predictive models.
The aim is to create a decision-aid support system that will enable fully automatic characterisation of the AAA morphology. It will also provide a function for surgical planning, providing automatic sizing of the vessels and an objective score of AAA outcomes, by integrating the risk of AAA expansion rupture and the risk of post-operative complications.
