Clinical Natural Language Processing and Health Interoperability to Support Knowledge Management and Governance in Rare Cancers

Rare cancers are a significant public health challenge due to their low incidence and complex nature, resulting in limited knowledge and resources for patient care and research. To address this issue, we present a Rare Cancer Data Ecosystem that enables the re-use of existing health data sources for rare cancers across European healthcare systems. Its main goal is to leverage emerging interoperability technologies and artificial intelligence approaches to improve the quality and organization of rare cancer patient care and advance health research. The ecosystem will be developed and tested through an European multidisciplinary project that will cover a regulatory-compliant data governance approach, a standard-based reference architecture, and a set of technical tools to share data, while ensuring data sovereignty policies. We aim to take part to the European Healthcare Data Space by contributing to an open, transparent digital ecosystem. Natural language processing and understanding tools will allow the extraction of health information stored in text form and support data findability and reusability. It will also include a secure pipeline for machine learning models development and deployment in a secure and reliable manner. The ecosystem will also include an augmented analytics and multimodal data navigator, providing a more natural and intuitive experience for users to explore and analyze rare cancer data. Overall, these features, after being piloted with several rare cancer data sources across different countries, will enable the efficient and secure use of existing health data sources for rare cancers, which can lead to improved patient care and health research.