Overview
Star date
1 January 2024
EU contribution
9.988.833,75€
End date
31 December 2027
Funded by
Horizon Europe programme
Duration
48 months
Project ID
101137074
Challenge
The HEREDITARY project addresses the critical challenge of leveraging multimodal health data to improve disease prevention, diagnosis, and treatment. Despite advancements in digitalization and machine learning, utilizing medical data remains complex due to its diverse nature—ranging from genomic data and bio-images to bio-signals and medical texts. Integrating this data from multiple sources is further complicated by technical and legal barriers, data heterogeneity, and privacy concerns.
To address these challenges, HEREDITARY targets three key objectives overcoming the state-of-the-art in multimodal health data integration, access and re-use.
Methodology
The HEREDITARY methodology is structured into five interconnected layers:
1. Federated Networking Infrastructure: This layer ensures secure and collaborative machine learning across different locations without transferring raw data. By using secure supercomputer environments, HEREDITARY maintains data privacy and complies with regulations like GDPR, enabling data-centric and model-centric federated learning.
2. Clinical Use Cases and Open Data: This layer focuses on collecting, preprocessing, and preparing clinical, genomic, and environmental data for federated analysis. Ensuring high-quality data, it facilitates robust clinical use cases and supports the discovery, querying, and training of federated learning models.
3. Multimodal Semantic Integration Platform: The platform integrates diverse data types using a polystore system, harmonizing access to public and private data. Employing Ontology-Based Data Access (OBDA), it enables seamless querying and analysis of clinical, genomic, imaging, and environmental data.
4. Multimodal Analytics and Learning Platform: This layer uses advanced AI and machine learning techniques to analyze multimodal data. It supports disease detection, treatment response preparation, risk factor identification, and evidence-based decision-making, handling the complexity and heterogeneity of biomedical data.
5. Visual Analytics and Interaction: Providing advanced visual analytics and interactive tools, this layer allows users to explore and analyze complex health data. It supports data exploration, hypothesis testing, and result presentation, enhancing transparency, explainability, and user engagement.
