AI & Data in Student Support: Practical Insights from the Netherlands and Belgium

How can we act before dropout becomes a reality? That was the guiding question behind our recent webinar on the role of AI and data in student support. With expert perspectives from Belgium and the Netherlands, we explored how institutions in both vocational and higher education are using predictive analytics, regional coordination, and explainable AI to identify students in need and to intervene before it is too late.

The conversation revealed not only promising innovations, but also the very real ethical, practical, and human questions that arise as we integrate technology into student guidance systems.

From Prediction to Personalisation in Flanders

Tom Madou, researcher at VIVES University of Applied Sciences and KU Leuven, shared how his team is using learning analytics and explainable AI to help students succeed in blended learning environments.

By analysing digital activity logs from their LMS and survey data, Tom’s team built predictive models that identify students at risk of underperforming early enough in the semester for intervention to be meaningful. Importantly, their models don’t just make predictions. They explain them. For example, a student might be flagged as at risk not simply because of low quiz scores, but because they haven’t visited discussion boards, which the model identifies as a success factor for that particular course.

Rather than replacing human insight, this approach enhances it. Tom emphasised that explainability is key: “We don’t just want to say, ‘You are at risk.’ We want to say why, and how you might act on it.”

Yet even well-performing AI models face challenges. GDPR constraints, fluctuating course content, and data drift between academic years all complicate implementation. Despite this, simple models often outperformed complex ones and yielded actionable insights for both educators and students.

A Regional Approach to Dropout Prevention in the Netherlands

Where Tom focused on predictive AI in higher education, Arno den Otter brought a regional lens from Dutch vocational education. As team leader at the Student Affairs Office for the Zuid-Holland Oost region, Arno coordinates one of 41 regional programmes across the Netherlands targeting early school leavers.

In the Dutch context, “early school leavers” refers to students who don’t complete their basic qualification, typically level 2 vocational education or pre-university pathways. These students are statistically more likely to face lifelong social and economic challenges.

Arno’s team uses a combination of school-based risk monitoring and municipal follow-up, including direct outreach via text or even home visits. Their interventions are guided by rich dashboards showing dropout hotspots at course and school level. Vocational programmes in logistics, for example, were identified as having higher dropout rates, prompting more targeted support.

While AI is not yet widely implemented, Arno sees potential: “We have the infrastructure, we have the data, and now we’re moving toward more predictive approaches. But we need to combine this with human care.”

Key Takeaways

Across both countries and approaches, several common themes emerged:

• Timeliness is everything. Whether through predictive models or real-time dashboards, early identification is essential for meaningful support.

• Explanation matters. AI models that offer transparent, understandable reasons for their predictions empower both students and educators to act.

• Human connection remains critical. Both speakers emphasised that data should lead to dialogue, not replace it. The goal is always to bring students closer to the right support—whether that’s a mentor, a care team, or a guidance counsellor.

• Data ethics and consent must be central. Transparency, student privacy, and informed consent are non-negotiable, especially when dealing with sensitive issues like mental health and dropout risk.
  

What’s Next?

At Annie, we are inspired by these approaches. Whether through rule-based bots or generative AI, our goal is the same: to reach the right students, at the right time, with the right kind of support.

As institutions across Europe explore how to use data and AI in student support, one thing is clear: the most powerful systems are those that balance technology with trust, prediction with care, and data with human empathy.

Interested in learning more about how AI is being applied in student support? Watch the full webinar recording here. 

Want to talk about how your institution could benefit from predictive support tools? Reach out to our team. We’d love to hear from you!