The National Institute for Health Research (NIHR) Cambridge Biomedical Research Centre (BRC) has supported the development of the world’s first artificial intelligence (AI) tool for Covid-19 treatment.

The study, titled EXAM, is said to be one of the “largest, most diverse clinical federated learning studies” to date.

This AI tool is built upon data collected across North and South America, Europe and Asia, and is used to predict how much extra oxygen a Covid-19 patient requires in the first few days of hospitalisation. 

The tool was tested in several hospitals globally, including Addenbrooke’s Hospital in Cambridge. It accurately predicted the amount of oxygen a patient needs within 24 hours of their arrival at the emergency department with a sensitivity of 95% and specificity of over 88%. 

To develop the tool, a machine-learning technique called “federated learning” was applied, using chest x-rays and electronic health data from over 10,000 Covid-19 patients.

Professor Fiona Gilbert, lead of the study in Cambridge, Honorary Consultant Radiologist at Addenbrooke's Hospital and Chair of Radiology at The University of Cambridge School of Clinical Medicine, said that “federated learning has transformative power to bring AI innovation to the clinical workflow”.

Dr Ittai Dayan, lead author on the study, stated that they developed the algorithm using federated learning. Dayan said that this allows for a “generalizable model” to be constructed, facilitating collaboration and data sharing between different hospitals across continents. 

Elaborating on the machine-learning technique's potential applications, Professor Gilbert continued: “The more we can securely integrate data from different sources using federated learning and collaboration, and have the space needed to innovate, the faster academics can make those transformative goals a reality.” 

NIHR Cambridge BRC, alongside Mass General Brigham and NVIDIA Inception startup Rhino Health, will continue to work on the EXAM Model and run prospective studies using the algorithm.