European Health Data Space – mobilising data for health innovation and AI development

Reuse and sharing of patient health data brings significant benefits for healthcare systems and individuals by making healthcare management more efficient, improving early detection of diseases and helping create safer and more effective medicines.  The European Health Data Space (EHDS) is a promising step towards facilitating the reuse of health data for such purposes in Europe. While the negotiations on the EHDS Regulation between the European Parliament and the Council of the EU are underway, it is worthwhile to keep in mind the advantages that data availability and reuse can give for research. 

When it comes to data, more is more for cure and AI

In the current era of data-intensive scientific research, broad data availability is a necessity. This extends not only to quantity but also quality and types of data. Nowadays, the classic registry health observations are often extended by numerous types of “omics” data that include large-scale molecular data like human genomic, gene expression, and structural data that can help to identify genetic alterations causing diseases, pinpoint exact causes of cellular malfunction, as well as help choosing individually tailored medications (read more: National Geographic and Genome Medicine). For full effect, as wide and open availability as possible of these data for multidisciplinary research is crucial. 
  
Broad data availability also facilitates the use of artificial intelligence (AI) in health research. AI is among the most promising digital technologies to advance care and research of human health with potential to recognise disease patterns and help making more accurate and faster diagnoses, in cancer for example. To fulfil this promise, AI must be fuelled with large amounts of good-quality data. Organisations prioritising public interest have a special role in developing such applications. It has been posited less data leads to fewer AI applications and data supply should be the focus, all the while the health AI market is dominated by the likes of Google and Meta and Europe is looking to build its own capabilities.

Accelerating the productivity of research and innovation could be the “most economically and socially valuable of all the uses AI”, the OECD highlighted recently. All valuable data sources need to be harnessed for this as part of developing responsible, widely adopted European AI. This is why it is a pity to see the Parliament propose limiting the scientific reuse of EHDS data only to health or care sectors. Yet, in the spirit of open science, the data made available with the EHDS could prove useful for AI development in a variety of fields, and more widely for entire academic disciplines like population genetics.

Striking a balance between smooth data reuse and strong data protection

Due to the sensitive nature of health data, a key part of the EHDS is individuals’ control over their own data. The EU co-legislators seem to take slightly differing positions on how to obtain individuals’ consent for the secondary use of their various health data: automatic consent with the possibility to opt out, or combining that to explicit consent/opt-in for genetic data. There are examples of large genomic data sets collected with explicit research use consent, but the opt-in approach is likely to significantly slow down wider adoption of genetic data use for European research that could improve public health. The Parliament’s request to promote public awareness of secondary use of electronic health data including the advantages and potential gains to science might have an impact in the long run, but only if the key aspects of the EHDS are done right.

One more way to increase data availability is to enhance people’s trust in the way it is processed. The EHDS only allows doing research on electronic data that is either anonymised or pseudonymised, and this in a secure processing environment (SPE) as an additional safeguard. These are useful ways to ensure privacy and data protection. It is also important to implement them in a way that does not create unnecessary hurdles for data reuse. For example, the upcoming SPE requirements by the Commission must strike the right balance between security and functionality. Additionally, pseudonymisation must be preferred over anonymisation whenever possible, as it allows a researcher to verify data, return research findings back to the source dataset and take different approaches, as well as communicate clinically significant findings about an individual’s health.

Investments in European data management and technologies benefitting citizens

The stricter the security requirements, the more resources the implementation of EHDS requires. Security-graded, privacy-by-design IT systems are expensive to produce for example due to the complexity of technical implementation of SPEs that both retain the utility, ease of use, and ensures the high level of security needed.  

For research and innovation, the most crucial aspect of EHDS will be quality and compatibility of research data made available from different EU member states. The initial focus should be on providing as comprehensive and clean datasets as possible and combine them with highly automated access procedures. Cleaning and harmonising data and developing the technologies is expensive and labour-intensive but will in the end benefit European citizens. Building on existing solutions and leaning on a more federated approach whenever possible would allow for a less expensive system and earlier availability.

Investments in digital technologies advancing European capabilities need to be prioritised when planning the remainder of the current Multi-annual Financial Framework and the forthcoming one, in line with the Parliament’s call for further resources for the EHDS. Although the full benefits of the EHDS might be years away, the building blocks of its success are defined now, and the opportunity must be capitalised on.

Authors:

Juhana Immonen
Policy Specialist at CSC

Heikki Lehväslaiho
Senior Application Specialist at CSC

Helena Lodenius
Project Coordinator at CSC