Data sharing, being the ultimate prerequisite for personalized healthcare, is hampered by various issues. PNA has investigated the various options and requirements for a Data Sharing Platform, that should alleviate these issues.

Personalized healthcare is a way of treating a patient based on the genetic characteristics of the disease and of the person themselves. Personalized healthcare has sparked the emergence of a multidisciplinary field combining genomics, healthcare and big data analytics. Healthcare professionals try to achieve this by combining data from various different sources and disciplines. The Personalized Healthcare Catalyst (PHC Catalyst) is an alliance of more than 40 organizations to accelerate the achievement of personalized healthcare in the Netherlands. The PHC Catalyst alliance tries to remove barriers and hurdles, by combining the joint power of all member organizations. Combining different disciplines and data from different sources is essential for realizing personalized healthcare. However:

  • The existence of data silos does not enable the consolidation or combination of different data sources, and the “right” data is either not stored or not stored in a homogeneous format. Data from different sources are often stored separately and in a wide range of formats. Hospitals, pharmaceutical companies, general practitioners, researchers and regulators all have their own data silo and even within one group, data are usually not stored in a homogeneous format. Discrepancies exist between the variables that are recorded. Hospitals store different information according to their own needs. There is not one standard to store the different types of data.
  • There is reluctance to share data outside the organization. Sharing data between organizations (and often also within organizations) is a sensitive topic.  Firstly, organizations fear the use of their data by others and are afraid to lose control. Legal frameworks do not fully cope with this discomfort. And secondly, data are a relatively new asset for  organizations. Most organizations tend to not yet fully understand how their data can be used. This makes it difficult to value data as an asset. If one cannot value their own data, it is even harder to value the data of others. As a consequence, it is hard to predict what the outcome will be when organizations share their data. Therefore, trust is a major factor in data sharing.

Data sharing platforms (DSPs) might play a role in alleviating these hurdles. PNA has conducted a project for the PHC Catalyst Alliance to investigate the feasibility and requirements for a DSP.

The study comprised of a literature review, an analysis of existing platforms, interviews with subject matter experts and a survey among members of the PHC Catalyst Alliance.

To enable an informed choice for a DSP, we developed a framework, consisting of three layers: scope, foundation, and configuration. This DSP framework is based on existing implementations of similar platforms, combined with a thorough review of academic literature and the consultation of a number of experts in the field.

The scope is set by making three assumptions on the implementation of the platform.

Within this scope we define three platform styles that act as the foundation for the sharing platform. The three styles are not to be seen in isolation but aspects of each can be combined to lay the foundation of the platform.

On top of the foundations we identify six parameters, based on review of existing platforms, the literature and interviews with experts. These parameters act as building blocks for configuration of the platform. Each parameter can be tailored towards the chosen platform style.

How did we populate this framework?

First and foremost, we found that control over shared data should always stay with the data provider, that access to the DSP should be on an invite-only basis, and that access to the metadata should be available to all DSP users. This defines our scope.

Our study shows, that the DSP should have the following characteristics, in terms of parameters and platform styles:

  1. The DSP has a centralized data store containing both unstructured and structured data vaults. In a later stage, the vaults can also distributed, i.e. located with the data providers. The vaults are, as indicated above, fully controlled by the providers of the content of the data vault.
  2. Standardization of data is facilitated through a glossary with clear and specific descriptions for the variables, containing strictly PHC related data. Governance of the standard is handled by a platform member or a third-party.
  3. Access to the platform is granted through invitation, whereas access to data is granted by data providers on a per-project basis.  Access to the platform takes place through a server with limited connections. Ownership of intellectual property should be determined by the involved members.
  4. Both a marketplace and a collaborative platform are feasible platform styles.

As a next step, we will develop a prototype based on this configuration, using building blocks that are available in the market. PHC is an exciting discipline, and a data sharing platform is one the infrastructural prerequisites to make it happen. The first results are here – there’s definitely more to come!

More on this project can be obtained by contacting Frank, at frank.harmsen@pna-group.com

More on the PHC Catalyst Alliance can be found here: https://www.phc-catalyst.nl/

Tomas, Lucas, Jos and Frank are consultants at PNA Group. Frank is professor Digital Transformation at Maastricht University.