What is the DRAGON project and why is it important?

COVID-19 hit the world hard. It impacted on everyone’s lives and overwhelmed healthcare resources worldwide, which lead to deaths that could have been prevented. On the individual level, it was difficult for people to know what was happening, how to prepare and behave, when to seek medical help and what to do post-COVID-19.

DRAGON is an Innovative Medicines Initiative (IMI) project which started on 1 October 2020, in the middle of the 2nd wave of the COVID-19 pandemic, and will run for 3 years. The project is coordinated by the University of Maastricht with Radiomics as the Project Lead. The partners include high-tech, small and medium sized enterprises, academic research institutes, biotechnology partners , together with the patient centred organisation, European Lung Foundation (ELF) and professional society, European Respiratory Society (ERS).

DRAGON is built on the ambition to increase the capacity of health systems, speed up the pace of research and innovation, and empower citizens during a pandemic if we:

  • improve how we diagnose people
  • are able to predict patient outcomes early on
  • empower citizens and patients to participate in their diagnosis as well as care and research

Because of progress in science and research DRAGON will use advanced approaches such as:

  • artificial intelligence (AI) – algorithms that learn and improve when they are provided with new information
  • patient and doctor decision support tools in the form of personal health applications (apps)
  • molecular profiling (a way of classifying a condition or disease outcome based of its genetic and other biomarkers)

These approaches will help to increase healthcare capacity and enable the research and innovation process by making them more efficient and provide patients and the public with tools to help and guide them during a pandemic.

What does DRAGON aim to do?

The aim of DRAGON is to increase the capacity of the healthcare system so it can respond to COVID-19 and to future pandemics.

To achieve this goal the project will achieve four main objectives:

  1. Deliver AI models that are based on imaging (such as CT scans) to diagnose and predict the course of COVID-19 and other infectious diseases. These models will support medical decision making and resource planning. These AI models will allow clinicians to diagnose and triage patients by providing a prediction of disease outcome within minutes. Certain features on medical images are impossible to fully assess with the naked eye which can be inaccurate. AI models are far more efficient and accurate than relying solely on the individual doctor.
  2. Speed-up the development of new therapies by creating a precision medicine approach. This approach will apply molecular profiling and AI to the models to diagnose and then predict the course of infection and response to individual treatment options.
  3. Use a machine learning system that will consider new information to continuously learn and improve. This “Distributed Learning” system, will provide a way of efficiently sharing and analysing data on a large scale. This will increase the capacity for innovation and the speed with which it can happen. By building this data sharing platform we can guarantee increased preparedness for future infection disease outbreaks and the ability to stop a future pandemic in its track before it reaches COVID-19 proportions.
  4. Engage stakeholders to develop tools to support personal decision making that will focus on empowering patients and the public. The support tool will consider the whole patient journey and will be built using information and learning from the first three objectives. Research shows that when we put the patient first and we empower individual patients to actively participate in their care, it leads to better patient outcomes overall.

What does this mean in real life?

The Distributed learning system will collect information about diagnosis and predicted outcomes from more than thousands of people and from different parts of the world. Using this system doctors will be able to input a person’s information from a medical examination and get a predicted diagnosis, recommended treatment and guidelines on follow up care. On the other hand, a patient app will enable them to understand more about the disease, advise them when to seek medical care, how to manage the condition and what to expect in the recovery phase. Together, the patient and doctor support tools can greatly improve and accelerate patient diagnosis and treatment. By learning by the example of the COVID-19 pandemic we can ensure that we are better prepared to cope with any future disease outbreaks and pandemics without overwhelming out healthcare system.

Example of a possible case study:

It is November 2022. The world has experienced nearly three years of the coronavirus pandemic. It is much less of a public health challenge with a global vaccination programme and moderately effective treatments.

A 75-year-old man comes into the emergency room at Maastricht University Medical Centre with shortness of breath. The doctor is concerned it might be a coronavirus infection, so the patient is put into isolation and examined. During the examination the patient explains that he has had the COVID-19 vaccine.

Once the man has been examined and had a CT scan and a chest X-Ray the doctor enters the information into the AI driven system. The information is compared with over 100,000 other examples that are stored on the system.

They wait a few minutes and the doctor checks the computer, the dashboard flashes red confirming that is has a “provisional diagnosis”, the doctor clicks on the message which says that the man may have coronavirus pneumonia.

The computer provides a protocol to deal with the situation and has alerted the public health authorities in the Netherlands along with the local infectious diseases’ expert. The doctor alerts his colleagues.

The system then recommends a treatment for the patient based on their examination which it has compared with other similar cases in the database. The patient is admitted to the intensive care unit.

Three days later the results of the patient’s sample testing come back and show that he has a novel coronavirus (not SARS-CoV-2). This places the public health authorities into full action, they trace his contacts and initiate a global response to a new coronavirus pandemic threat. The learning system is updated with these results, which are the first data to be used to help combat the new coronavirus.