Newsletter #3 Hello,
as IDEA4RC enters its second year, consortium members gathered in Madrid to discuss the steps ahead and the key milestones for the second year. It was the occasion to outline the activities that will lead to the first infrastructure deployment within the 11 clinical centers, and kick-off the activities regarding natural language models, which will be developed to extract relevant information from textual documents, such as medical notes and pathology reports. In addition, partners started discussing possible governance approaches, and the functionalities of the virtual assistant, which will allow researchers and clinicians to interact with the data ecosystem.
The last two months were crucial for the definition of the legal landscape in which IDEA4RC will fit. The Data Act was adopted by the Council of the EU, and a provisional agreement was reached between the Council and the European Parliament on the AI Act. Finally, the Council and the Parliament have formally received mandate to start negotiations on the European Health Data Space. You can find a summary of these news in the last section of this issue.
If you missed the first two issues of this newsletter, where we introduced the project, its motivations and objectives, and wrapped up the first year of the project you can read them here.
By subscribing to the newsletter you will be receiving bi-monthly updates on the projects advancements. If you want to invite your friends to subscribe, send them this link.
What to expect during the second year of IDEA4RC
The IDEA4RC third consortium meeting was held at the Universidad Politécnica de Madrid on 22-23 November 2023, hosted by the Life Supporting Technologies lab. The second year of the project is for health professionals and providers, said project coordinator Annalisa Trama in her opening remarks. We really need your contribution to achieve the milestones planned for the end of this second year.
The milestones include:
- infrastructure deployment, with 11 clinical centers testing the capsule environment using synthetic data;
- data ingestion, where clinical centers locate variables in the common data model and identify those requiring Natural Language Processing (NLP);
- and the establishment of the governance and legal frameworks.
Three working groups, focused on architecture, NLP training and testing, and pilot deployment, have been set up to meet these milestones. The pilot deployment group, involving health professionals, hospital data managers, and technical partners, ensures clinical centers stay updated and aligned with the projects timeline.
In addition, during the second year, specifications for the functionalities and users of the IDEA4RC virtual assistant (VA) will be finalized, considering the information gathered from focus groups and interviews with clinicians and researchers. The VA will be the interface between users and the IDEA4RC data ecosystem. Preliminary results on VA specifications were shared with the clinicians during the meeting in Madrid and led to an insightful discussion.
You can read a more detailed account of the meeting here.
Meetings, results and updates
- Claudia Egher of Utrecht University led the second IDEA4RC co-creation workshop in Madrid with nearly 50 participants from consortium partners. The workshop aimed to discuss the three possible implementation scenarios envisaged for IDEA4RC platform. Participants explored the desirable end situations and societal values for each scenario. You can read a summary of the workshop here.
- IDEA4RC was presented at the 5th International Conference on Rare Diseases, which was held in Budapest on 30 November and 1 December. You can read about the conference and have a look at the poster here.
- IDEA4RC members will take part in the 35th Connectathon organized by the HL7 FHIR community on 16-18 January (online). IDEA4RC will be taken as a reference example in the European Cancer Mission track of the Connectathon, which main aim is to compare existing European FHIR Cancer-related scenarios and Implementation Guides to identified gaps and achieve some levels of alignment. Further information is available here.
Whats up in health data sharing and reuse in the EU
- On 23 November, the Council of the European Union adopted the Data Act, a new Regulation on harmonised rules on fair access to and use of data. The Data Act places obligations on manufacturers and service providers requiring them to enable their users, companies, or individuals to access and reuse the data generated by the use of their products or services. Furthermore, it grants users the ability to share this data with third parties. As a result, consumers gain the flexibility to easily switch between different cloud providers. Additionally, the regulation empowers public sector bodies to access and utilize data held by the private sector, especially in exceptional circumstances such as public emergencies (e.g., floods or wildfires) or to fulfill tasks in the public interest. Following the Data Governance Act adopted in 2022, the Data Act is the second main legislative initiative within the European strategy for data launched by the European Commission in February 2020. For a more comprehensive understanding, you can access the complete press release from the Council of the European Union here.
- On 9 December, the Council of the European Union and the European Parliament reached a provisional agreement on the AI Act following three days of negotiations. This act is designed to regulate the use of artificial intelligence algorithms within the European Union, placing a strong emphasis on prioritizing safety, respecting fundamental rights and EU values. The regulation employs a risk-based approach, categorizing AI applications into three risk levels. Limited-risk applications face lighter transparency obligations, while high-risk AI applications?such as those used in hiring and education?are subject to more stringent rules. Developers of high-risk tools must provide evidence of risk assessments, disclose data used for training, ensure human oversight, and prevent harm, particularly in perpetuating biases. Certain uses of AI deemed as posing unacceptable risks will be banned from the EU. These include cognitive manipulation, facial image scraping, emotion recognition, social scoring, biometric categorization for sensitive data, and specific predictive policing cases. Exceptions are made for real-time biometric surveillance for national security reasons. Foundational models and general-purpose AI systems must comply with transparency obligations before reaching the market, with a stricter regime for high-impact models. It is the first legislative proposal of its kind in the world, and could thus set a global standard for AI regulation. The agreed-upon text must now undergo formal adoption by both the Parliament and Council to become EU law. For more details, you can access the complete press release from the European Parliament here.
- The Council of the European Union and the European Parliament have officially received mandate to start negotiations on the European Health Data Space (EHDS). Each institution adopted amendments to the draft regulation proposed in May 2022 by the European Commission. These documents will constitute the ground of the forthcoming negotiations. The EHDS aims to facilitate the accessibility and exchange of health data across borders. This supports healthcare delivery (primary use of data) and contributes to health research and policy-making (reuse of data, also referred to as secondary use of data).
Regarding the primary use of health data, the Parliament expanded the proposal of the Commission by introducing the possibility for patients to restrict access to anyone except the health professional who inserted the electronic health data. The Parliament also introduced an amendment to allow patients to opt out of the processing of their electronic health data for secondary use. A stricter position has been taken on genomic and genetic information, with the Parliament proposing an opt in system. This means that every patient whose data is involved will have to give consent each time this data is used.
Instead, the mandate of the Council requires the establishment of two steering groups, made up of member-state representatives, to manage MyHealth@EU and HealthData@EU, the two pilot infrastructures that enable the exchange of health data across member states for primary and secondary use, respectively. The mandate also expands the role of EU member states in the proposed EHDS governing board and requires national digital health authorities to publish an activity report every two years. Additionally, member states will have the discretion to allow patients to opt out of the new data-sharing system.
Thats all for this issue. See you in February with an interview to the consortium legal experts about the implication of the Data Act and AI Act for health data re-use in general and for the IDEA4RC platform in particular.
THE IDEA4RC consortium
|