Skip to main content
European Commission logo print header

The RapiD and SecuRe AI enhAnced DiaGnosis, Precision Medicine and Patient EmpOwerment Centered Decision Support System for Coronavirus PaNdemics

Periodic Reporting for period 2 - DRAGON (The RapiD and SecuRe AI enhAnced DiaGnosis, Precision Medicine and Patient EmpOwerment Centered Decision Support System for Coronavirus PaNdemics)

Reporting period: 2021-10-01 to 2022-09-30

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. We 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.
The project objectives are to:
1. Deliver scalable diagnostic and prognostic models based on imaging that are more efficient and accurate for supporting medical decision making and resource planning.
2. Accelerate new therapy development by developing a precision medicine approach that adds molecular profiling and AI enhanced analysis to the multi-faceted scalable diagnostic and prognostic models.
3. Deploy a federated machine learning system that will support fast track innovation by enabling continued data driven improvement while expanding the innovation capacity of this and other initiatives by providing a means to efficiently share and analyse data at scale.
4. Engage stakeholders in the development of a patient empowerment centered decision support system that considers the entire patient journey and incorporates the outputs of the first three objectives.

The successful achievement of the objectives of this project will provide a number of global benefits:
1. Rapid accurate diagnosis in patients with suspected viral pneumonia, facilitating the timely implementation of isolation procedures and early intervention.
2. Increased prognostic precision allowing optimum resource management (including management decisions regarding the allocation of ventilators)
3. Development of cohort enrichment strategies for treatment trials, therefore accelerating drug development.
4. Rapid development of a federated network, ensuring that connected healthcare systems are better prepared for future pandemics
Four main objectives of the DRAGON project were identified. From month 12 to month 24, the DRAGON consortium was able to address the majority of the objectives defined in the Description of Action:

i. Deliver scalable diagnostic and prognostic models based on imaging

During the last and this reporting period, the consortium was able to develop diagnostic models, but with the rapid evolution of the pandemic and the slow Medical Device Regulation (MDR) approval processes, these models are no longer feasible to answer clinical needs. With this, it was decided to shift the focus from diagnostic models to prognostic models that can determine which patients might have signs of long-COVID, which patients require a more thorough follow-up and how to better implement precision medicine approaches. The work here is closely related and dependent on data collection and its availability via the distributed learning network set up. Clinical partners have been collecting and curating data, with the guidance of CDISC standards to get the data machine readable. Some partners were able to successfully finalize this step and are uploading data into the DistriM machines. Further actions to fully set-up the distributed learning network have been taken, such as the preparation of mock data and initial development of potential prognostic models.

ii. Accelerate new therapy development

To enable precision medicine approaches to patient care, a subset of patients who are at risk of severe disease or death were identified so that they can receive enhanced treatment. Partner University of Maastricht has made great strides in this respect, having developed a model that predicts the risk of severe disease during hospitalization that was published in August 2020. It used age and sex features from routine blood tests. That model is meant to be prospectively validated using data from multiple clinical partners (in and outside the DRAGON consortium). Additionally, they have delivered a knowledge graph based on existing knowledge and a report on multi-factorial analysis of existing data, the associated work led to four peer-reviewed publications. In the meantime, the consortium led a multi-factorial analysis with the data collected within the DRAGON consortium.

iii. Deploy a federated machine learning system

Partner Oncoradiomics has gathered input from clinical sites on the patient data available and together with CDISC developed a curation strategy for this data. Input from other AI developers in the consortium was crucial to shape some of the requirements for the distributed learning software in development (C-DistriM). Data upload tests were performed and first efforts to officially upload data into DistriM started in September 2022. Interactive activities took place to engage model developers with framework developers to successfully deploy the federated machine learning system once the data is fully available from all clinical partners. DRAGON consortium also identified potential European initiatives to potentially start side collaborations to integrate the distributed learning network.

iv. Engage stakeholders in the development

We focused on developing communities of stakeholders and engaging with them in different ways (e.g. conferences, focused groups, test rounds, roleplay roundtables) to provide important insights on their respective needs and perspective on a regular basis. The DRAGON consortium took most into consideration patient involvement to address this objective. These insights were important to steer the consortium towards the development of relevant tools and models and increase adoption rates. Based on these stakeholders' interactions, partner Comunicare Solutions designed an app (https://www.comunicare.be/home/) for the information and monitoring of COVID-19 patients and is in close contact with other WPs to identify outputs to be used within a Decision Support System.
Despite the numerous warnings of the potential catastrophic effects of an unexpected and novel pathogenic outbreak, the COVID-19 pandemic has revealed that additional innovative steps, beyond simply expanding capacity, resources and coordination, are needed to increase our readiness for and effectiveness during such an infectious disease event. With the aim of upgrading current and future pandemic response capabilities, the application here will integrate AI enhanced imaging and distributed data analysis approaches to develop a universal infrastructure that will:
● Validate the concept that advanced AI techniques extract the central insights from CXR and CT scans to facilitate rapid diagnosis and prognosis.
● Establish proof-of-concept for the use of AI enhanced precision medicine approaches to improve prognosis and accelerate the development of new therapeutics.
● Increase preparedness for individual case management with patient empowerment centered decision support systems.
dragon-visual-abstract.png