Niels Greidanus became CEO of CODE24 at the beginning of 2025 and was the driving force behind the launch of Cadasto in January 2026. While he can rely on years of experience in healthcare IT, the world of open standards was relatively new to him – though joining CODE24 provided a rapid introduction to its possibilities. An entrepreneur to the core, Niels follows developments concerning AI with keen interest. Even so, he recognises that using AI in a healthcare context presents unique challenges, where structured data modeled according to openEHR has a crucial role to play. In this article, we share his perspective.
Reliability Is Key
When it comes to using AI in healthcare, there is no margin for error. AI solutions need to be reliable and consistent. “With AI, the quality of the output is always reliant on what you put into it”, Niels explains. “Garbage in means garbage out, as they say. So you need to make sure that the data you are ‘feeding’ the AI is of the highest quality and structured in such a way that it leaves the AI minimal room for error.”
So, how should that data be structured? “With a standard like openEHR, data is modelled according to a predefined structure”, Niels enthuses. “This means that the data is easily computable, consistent and rich in context. This is exactly what you need to safely use AI in healthcare.”
One Data Source
With open data platforms such as Cadasto, IT landscapes are increasingly moving towards an architecture that separates data from applications. This is good news for organisations who are looking to embrace AI. While care organizations have historically struggled with fragmented IT landscapes, decoupling data from applications creates a single, unified source of data for multiple systems.
“The more context AI has, the better the output”, says Niels. “When data is stored in Cadasto and reusable across applications, there is only one data source for AI to query. This means you no longer need to extract data from multiple silos or run complex conversions before putting it to work.”
Start Sooner Rather than Later
This sounds ideal – but Niels is also realistic. “Migrating entire IT landscapes to an openEHR platform takes time. Even when you do, historical data is usually based on proprietary models and there will be gaps in the context. openEHR is very comprehensive – but certainly not all models are. Data conversion will only take you so far.”
Even so, this only means that it is important to start sooner rather than later. “There is clear consensus on where healthcare IT must head regarding data. Regulation is increasingly pushing for the separation of data and applications, and the necessity of structured data is clear. The market needs to move – but it is slow-going. To truly leverage data-driven processes and AI tooling, we need to accelerate. The best path forward is to start with a small, focused pilot, learn from the results, and scale up step by step.”
Ready to get started?
Would you like to discover what our open data platform has to offer?


