Rapid and secure AI imaging based diagnosis, stratification, follow-up, and preparedness for coronavirus pandemics


Start Date
End Date
IMI2 - Call 21
Grant agreement number

Type of Action: 
RIA (Research and Innovation Action)

IMI Funding
11 381 970
EFPIA in kind
160 672
Total Cost
11 542 642


Diagnosing COVID-19 and, crucially, predicting how the disease will progress in different patients, remains a challenge. The DRAGON project aims to use artificial intelligence (AI) and machine learning to develop a decision support system capable of delivering a more precise coronavirus diagnosis and more accurate predictions of patient outcomes.

The project will draw on new and existing data and sample collection efforts, including CT (computed tomography) scans to carry out detailed profiling of patients. They will then use AI technology to transform this information into a precision medicine approach that will help clinicians and patients with decision making around treatments.

Underpinning all of this will be a federated machine learning system that will allow the use of data from a range of international sources while complying with the EU’s General Data Protection Regulation (GDPR).

A patient and public advisory group will provide advice and input throughout the project.

Achievements & News

To quarantine or not to quarantine? Predictive models can guide doctors’ judgement
September 2021

Around the world, doctors are using artificial intelligence to help them make decisions about patient care during the COVID crisis. ###Feeding on clinical, laboratory, genetic, and radiological datasets, machine learning models are able to churn out predictions that can be used to identify, for example, who ought to self-quarantine, and who ought to make their way to the hospital.

With more and more of these models popping up around the globe, the DRAGON project set out to create an online platform that would serve as an open source repository for a curated subset, with a simple interface that allows users to make online calculations. The website can be used by doctors to supplement their judgment with patient-specific predictions from externally validated models in a user-friendly format.

DRAGON sought out publicly available, validated, peer reviewed or open-source models and published them alongside supporting documentation and links to associated articles. The platform is dynamic and growing; it currently features nine models, and will continue to be populated with others as they become available. It is hoped that the platform will help speed up the adoption of predictive models, moving them from the research world into clinical practice.

Find out more

Study: face masks cannot replace swabs for testing COVID19 viral load
July 2021

A team that included researchers from the DRAGON project has found that viral load is lower on face mask filters than on nasopharyngeal swabs. ### Face masks and personal respirators can help stop the spread of droplets that carry the SARS-CoV-2 virus, and while nasopharyngeal swabs are the most dominant method for the collection of samples for COVID-19 diagnosis, the DRAGON project wanted to know if filters embedded in this personal protective equipment could be used as a non-invasive way to collect samples for, say, at-home testing.

DRAGON conducted a study where they generated inactivated virus-laden aerosols and dispersed them onto filters within face masks. These laboratory-based tests could detect coronaviruses down to a level of 10 copies per filter. However, testing of around 45 clinical samples suggested that the viral load emitted in breath aerosols of most patients with COVID-19 fell below this threshold. The difference in detection of SARS-CoV-2 between filters and nasopharyngeal swabs suggests that the number of viral particles collected on the face mask filter was below the limit of detection for all patients except those with the highest viral load – which has been shown to peak just before the onset of symptoms. This indicates that face masks are unsuitable for replacing nasopharyngeal swabs in the diagnosis of COVID-19. However, it might yet be suitable for chronic infectious agents where pathogen release is sustained, rather than having a transient period of high emission followed by a rapid resolution.

Find out more

Participants Show participants on map

EFPIA companies
  • Owlstone Medical Limited, Cambridge, United Kingdom
Universities, research organisations, public bodies, non-profit groups
  • Centre Hospitalier Universitaire De Liege, Liege, Belgium
  • Department of Health, Leeds, United Kingdom
  • Imperial College Of Science Technology And Medicine, London, United Kingdom
  • Lungs Europe, Brussels, Belgium
  • The University Of Liverpool, Liverpool, United Kingdom
  • Universita Degli Studi Di Firenze, Florence, Italy
  • Universita Degli Studi Di Parma, Parma, Italy
  • Universiteit Maastricht, Maastricht, Netherlands
  • University Of Southampton, Southampton, United Kingdom
  • University of Cambridge, Cambridge, United Kingdom
Small and medium-sized enterprises (SMEs) and mid-sized companies (<€500 m turnover)
  • Biosci Consulting Bvba, Maasmechelen, Belgium
  • Cdisc Europe Foundation Fondation, Brussels, Belgium
  • Comunicare Solutions, Seraing, Belgium
  • European Respiratory Society, Lausanne, Switzerland
  • Oncoradiomics, Liege, Belgium
  • Thirona BV, Nijmegen, Netherlands
  • Topmd Precision Medicine LTD, Southampton, United Kingdom
Patient organisations
  • European Lung Foundation, Sheffield, United Kingdom
Project coordinator
Philippe Lambin
Universiteit Maastricht