Making research data FAIR… How do you do that?

Organising your data from dementia research in line with the FAIR principles is so important – but it also turns out to be tricky. Why is that? And how do you manage it anyway?

Ideally, you should make your research data FAIR: findable, accessible, interoperable and reusable. That’s how you make the greatest impact, both within the scientific community and beyond (see ‘FAIR data: what is it (and what isn’t)?’). That’s why funding bodies often make FAIR data a requirement.

But is it actually possible to organise your data in a FAIR way? It varies, according to a survey of dementia researchers conducted by Dr Didi Lamers, data management coordinator at DEMPACT. Lamers identified the obstacles and how to overcome them.

“Making your research data discoverable and accessible – that’s still doable. But making it interoperable and reusable is a lot trickier in practice.”
Lamers is keen to share four key obstacles and (fortunately) the paths towards solutions.

Obstacle 1: The complexity of legal agreements

They are tricky, unavoidable and useful: the legal agreements you need to put in place to be allowed to exchange data with fellow researchers from other research organisations. Lamers heard about two major obstacles in this regard.

Slow process

Firstly, it proves difficult to reach legal agreements with all parties before you can even begin exchanging data with other research institutions. This is partly because lawyers from different research institutions have differing views on what should be included in those agreements. Communication on this matter can also be difficult at times. All in all, the coordination process takes a long time.

Legal versus scientific purpose

The aim of the agreements is to create a situation in which you, as a researcher, can exchange data legally, efficiently and with mutual trust. However, legal advisers sometimes focus on covering all risks (even unlikely ones). Within the finalised agreements, it may then turn out, for example, that sharing or publishing data is not permitted at all. Or the agreements have become so complex that they are difficult for non-lawyers to understand. In that case, the agreements miss the mark (building trust).

Obstacle 2: Yes, but what if…?

In discussions about FAIR data, Lamers also noticed that some researchers feel somewhat vulnerable about making their data accessible. What if someone else re-analyses it and reaches different conclusions? What if someone misinterprets your data? And what if your data is misused in other dementia studies? It’s perfectly understandable. It might feel as though you’re losing control over your data. But that doesn’t have to be the case at all.

You can organise your data in such a way as to minimise the risk of negative consequences. For example, by providing context: make sure you add rich metadata to help fellow researchers understand your data properly.

And make active contact: include contact details in your data publication. That way, fellow researchers who wish to use your data can contact you with any questions. Use an email address that will remain active for a long time. Is misuse of data harmful or ethically objectionable to your research participants? If so, you can restrict access to your data. This allows you to retain control. You assess every request for data access before granting access.

Obstacle 3: Researchers are reinventing the wheel

However difficult it may be to organise dementia data in accordance with the FAIR principles, solutions already exist for various issues – but where can you find them? This turns out to be a problem. Scientific publications are of limited use for sharing practical knowledge about FAIR data. There is currently no alternative source. Furthermore, whilst many dementia researchers are keen to exchange knowledge and experience, they are unable to find the right colleagues. DEMPACT can help solve both problems.

Zenodo: Dutch Dementia Data Community

DEMPACT will ensure that knowledge and solutions relating to FAIR data that are relevant to dementia research can be found in one place: the leading Open Science digital research repository, Zenodo. Lamers has set up the Dutch Dementia Data Community there.

You can find the publications listed on zenodo.org.

Insight into FAIR solutions

DEMPACT has a comprehensive overview of FAIR data solutions both within and outside the Dutch dementia sector, as well as the experts involved. These include data harmonisation, setting up federated analysis and other areas of expertise.

• Do you have a question or are you facing a challenge? Didi Lamers can put you in touch with relevant experts.
• Have you developed (or are you using) a solution that you’d like to share more widely? Please get in touch.

Obstacle 4: Lack of standards

If you wish to use data from multiple sources for your dementia research, it is helpful if measurements have been carried out in a standardised manner. Think, for example, of cognitive tests or the measurement of biomarkers. Data must also be documented and stored in accordance with a standard. If one researcher uses the code F for women and M for men, whilst another uses 1 for women and 0 for men, it becomes difficult to analyse the data coherently, particularly if there is no (complete) code book available.

Analysis is then only possible once the data from different sources have first been harmonised. This means that data are presented in a comparable manner.

Step one towards standardisation

DEMPACT aims to contribute to standardisation. Lamers will be organising knowledge-sharing and expert sessions where you can learn from the experiences of fellow researchers regarding making dementia data FAIR. This is the first step towards reaching a consensus within Dutch dementia research on which standard we will use.

"FAIR data covers a lot of ground. If you want to get everything perfect straight away, you’re probably setting the bar too high," emphasises Lamers. "You’ll make the greatest gains simply by making your data discoverable in the first place, via a repository or catalogue and with clear metadata. Even if your data aren’t yet optimally exchangeable or reusable, they can still be useful to other researchers.”