Unleashing the power of Artificial Intelligence
An open source data hub for haematological diseases
GenoMed4All goes beyond the state-of-the-art to diagnose, treat and predict haematological diseases by adopting groundbreaking and trustworthy AI algorithms. Our precision medicine model will feature Federated Learning and High-Performance Computing (HPC) to become more reliable when simulating and predicting patterns that might reveal weaknesses and risks.
Our core principles
INNOVATION
Using Graph Neural Networks to distributedly train deep learning algorithms
INTEROPERABILITY
Enabling collaborative, cross-border data sharing that is standard-compliant
TRUST
Offering security and privacy by design in all data exchanges, model training and storage
SCALABILITY
Using containerization to build modularity and facilitate massive replicability
What are we aiming for?
86+
Clinical repositories with genomics data connected across 15 EU countries
15%
Accuracy improvement in specific genomic markers for prognosis and treatment
80%
Increase in the adoption of open standards for -omics data per clinical site
20%
Time reduction in AI analysis and model training through Supercomputing
Linking -omics and clinical data repositories across Europe
The project will build a large-scale distributed repository of -omics health data across Europe, including: Electronic Health Record, PET, MRI and CT, Next Generation Sequencing, Microarray, Genome-Wide Association, Copy Number Variations, DNA and RNA sequencing. This scheme will enable the aggregation of a high number of repositories that are currently dispersed and non-homogenised while respecting the patient's rights
More accurate Deep Neural Networks regression and classification models
Explicit features extraction using advanced generative models such as Variational Autoencoders and Generative Adversarial Networks
Optimal fusion architectures using heterogeneous data as a combination of feedforward, convolutional and recurrent networks
Exploring new models in genomics for personalised medicine
GenoMed4All will deploy 'white box' AI models in 3 real-world pilots, improving the diagnosis and prognosis capacity, evaluating treatment options and patient response. The pilots will cover common and rare oncological (Myelodysplastic syndromes and Multiple Myeloma) and non-oncological (Sickle Cell Disease) haematological diseases
AI- based services for clinical support
DIAGNOSIS
AI algorithms for early identification of high-risk individuals
PROGNOSIS
Prediction algorithms for insights on disease development
TREATMENT
Clinical algorithms to aid decision-making in risk stratification of personalized therapies
The EU dimension
Transforming Health and Care in the Digital Single Market
EU policies on eHealth aim to create a European health data space to foster targeted research, diagnosis and treatment, ensure access to safe and top quality digital services in healthcare and pave the way towards a healthier society.
Priorities
ENABLING
Secure access to citizen’s health data across the EU
BUILDING
A shared EU data infrastructure for precision medicine
EMPOWERING
Citizens with digital tools for feedback and person-centred care