Silvia Belotto
At the LabWare 2026 European Customer Education Conference in The Hague, Theo Boersma from Wetterskip Fryslan and Karel de Jonge from LabWare delivered an insightful session titled “Applicability of AI for Laboratory Analysis.” Their presentation explored how laboratory data, governance, and advanced analytics can be brought together to support artificial intelligence initiatives in a regulated environment. Using a real-world water quality use case, they demonstrated how LabWare’s Data Science Engine can help laboratories transform historical data into actionable insights while maintaining transparency, validation, and human oversight.
Theo Boersma and Karel de Jonge at the LabWare 2026 European Customer Education Conference.
Building a framework for successful AI adoption
Artificial intelligence is generating significant interest across laboratory environments, but successful adoption requires more than access to powerful algorithms. The quality of the underlying data, the governance framework surrounding its use, and the ability to validate results all play a critical role in determining whether an AI initiative delivers meaningful value.
For Wetterskip Fryslan, the regional water authority responsible for water quality management in the northern Netherlands, years of historical analytical data created an opportunity to investigate how AI models could support laboratory decision-making and reveal patterns that would be difficult to identify through traditional analysis alone.
However, technology cannot be separated from governance. Regulatory expectations, security requirements, data integrity principles, and organizational policies must remain central to any AI strategy. Rather than treating AI as an independent decision-maker, Wetterskip Fryslan views it as a tool that supports laboratory experts by providing additional insight while keeping accountability firmly in human hands.
This balanced approach helps ensure that AI outputs remain transparent, understandable, and suitable for use in a controlled laboratory environment where trust in results remains essential.
Turning historical laboratory data into predictive insight
To explore the practical application of AI, the team selected nitrogen analysis as an initial use case. Historical datasets from multiple water-related contexts, including wastewater and surface water analysis, were connected through the LabWare Data Science Engine and used to train, validate, and test a range of machine learning models.
The objective was not simply to generate predictions, but to evaluate whether AI could provide an additional layer of understanding for laboratory data. By analyzing large volumes of historical information, the models were able to identify trends, relationships, and potential future outcomes that could support more informed decision-making.
The project also highlighted an important reality of laboratory AI: there is rarely a single model that works best for every situation. Different analytical scenarios may require different approaches depending on the quality, structure, and complexity of the available data. Seasonality, variability, sample characteristics, and data volume can all influence performance.
By systematically evaluating different modelling approaches, Wetterskip Fryslan was able to develop a clearer understanding of where AI could provide value and where traditional scientific expertise remained essential.
Keeping humans in the loop
“AI should not replace expertise, it should support experts by accelerating insights, identifying trends earlier, and improving decision-making.” – Theo Boersma
This principle sat at the heart of Wetterskip Fryslan’s approach to AI. While machine learning models can analyse large volumes of laboratory data far more quickly than a human analyst, scientific judgement remains essential for interpreting results, assessing risk, and determining whether recommendations are appropriate within the wider operational context.
For that reason, AI was presented as an additional analytical tool, fully under control of human oversight, which remains the central part of the process, helping to ensure that results are validated, understood, and used appropriately.
The project also highlighted several practical lessons for laboratories considering their own AI initiatives. Starting with a clearly defined use case, working with high-quality datasets, maintaining robust validation processes, and ensuring transparency throughout development can significantly improve the likelihood of success. Just as importantly, organisations should avoid treating AI as a black box. Understanding how models are built, tested, and monitored is essential for maintaining confidence in the results they produce.
Looking ahead
Wetterskip Fryslan’s experience demonstrates that effective laboratory AI is built on far more than algorithms alone. Success depends on a combination of reliable laboratory data, strong governance, robust validation, and continued human oversight. By bringing these elements together through platforms such as LabWare’s Data Science Engine, laboratories can begin transforming historical data into actionable insight while maintaining confidence in the results.
Ready to unlock the value of your laboratory data?
Discover how LabWare’s Data Science Engine, LIMS and ELN solutions can help your organisation explore AI-driven insights while maintaining data integrity, transparency, and regulatory compliance.
Learn how a strong data foundation can support future AI initiatives and help accelerate laboratory innovation: request a demo now.