Health data exists in many forms and multiple fragmented repositories; there is still significant room for improvement in the way both structured and unstructured health data is stored, analysed and interpreted. Sharing and analysing data from multiple countries in a safe and legally compliant manner (in particular with regard to personal data protection) remains a challenge. Powerful analytic tools are already helping providers to use structured data in increasingly impactful ways. On the other hand, the heterogeneity, diversity of sources, quality of data and various representations of unstructured data in health care increase the number of challenges as compared to structured data.
Advances in AI and machine learning, however, have the potential to transform the way clinicians, providers and researchers use unstructured data. Furthermore, developing data interoperability standards, trust and harmonization of GDPR’s interpretation across the EU for the sharing and processing of personal health data will support establishing a sound health data culture in view of the European Health Data Space.
Proposals should address all of the following aspects:
- Developing robust novel solutions compliant with legal requirements (in particular concerning personal data protection) that will improve the quality, interoperability, machine-readability and re-use of health data and metadata in compliance with FAIR data management principles, making these data more accessible to clinicians, researchers and citizens. The focus should be on data in electronic health records (EHRs) and/or patient registries, taking into account the Commission Recommendation on a European Electronic Health Record exchange format[[https://ec.europa.eu/digital-single-market/en/news/recommendation-european-electronic-health-record-exchange-format]].
- Developing innovative natural language processing tools, including computational semantics, ontologies, text mining, associated machine learning and deep learning, to improve accessibility, interoperability, translation, transcription, and analysis of health data (e.g. to predict risks). Tools should extract health information from unstructured data in different clinical and medical sources, and bring that data into EHRs/patient registries in a structured form. The innovative solutions should also address missing data in EHRs and/or patient registries and their related metadata, to reduce bias and improve the quality of conclusions.
- Developing and piloting AI-powered virtual assistants that will utilise the tools and solutions developed (as mentioned above) in order to demonstrate improved usability of health data for end-users.
Proposals are expected to build on and contribute to existing European and international data standards, specifications and schemas for health data. The use of open standards should be considered and interactions with relevant ongoing research infrastructure efforts are encouraged. Applicants should focus on health data coming from a number of EU Member States and EEA countries, constituting as much as possible a representative sample of the European healthcare landscape, so as to contribute to the work on the creation of the European Health Data Space.
To guarantee their adoption, the developed solutions should be quick and easy to use by researchers and clinicians. Therefore active involvement of end-users from the onset is encouraged. In particular, patient advocacy groups and citizens should be involved to ensure adequate consideration of diverse patient needs, with respect to their gender, ethnicity, age, ability, and socio-economic background, to underpin acceptance by patients and other data subjects. SMEs participation is also encouraged.
The proposals should duly take into account requirements stipulated in the relevant European regulations (Data protection, in vitro diagnostics and medical devices) and must meet appropriate ethical standards.