Impact |AI-assisted communication for people with aphasia

In the coming weeks the nominees for the Ben Feringa Impact Award 2026 will introduce themselves and their impactful research or project. The winners will be announced on 9 June. This week: Thijs van der Laan with the research on AI-assisted communication for people with aphasia.

Who are you?

Thijs van der Laan, graduated in Artificial Intelligence. What drew me to this field was the'newness' around it and its broad range, combining various disciplines. With a broad interest in various fields, I enjoy tackling complex and innovative problems, ideally applying AI tools and techniques to solve real-world challenges. I believe there are significant opportunities for positive impact in the application of artificial intelligence creatively across domains, such as aphasia. Currently, I work as a trainee at a financial institution, where I focus on ensuring responsible adoption and governance of AI. I studied at the Faculty of Science and Engineering, where I completed an MSc in Artificial Intelligence.

Can you explain what your research was about?

My research lays the foundation for studying language models as assistive technology for people with anomic aphasia, a condition that makes it difficult to retrieve and speak words. Their speech is characterized by frequent pauses or substitutions (paraphasias). I investigated whether language models can fill in these pauses or substitutions by drawing on the preceding textual context.

I built and evaluated an end-to-end pipeline that predicts likely next words at points of hesitation. For this, I used interview data from AphasiaBank, a large international dataset. Since my research envisions a potential app or smartwatch tool, I reduced the model sizes, modified the training function to improve prediction accuracy, and examined the strengths and limitations of this approach.

What made the research impactful?

People with anomic aphasia struggle in everyday conversation. Existing clinical tools offer limited support because they typically rely on fixed word or picture sets rather than natural speech, which often leads to exhaustion and social isolation. My research lays the foundation for AI-powered communication tools. Because AI models encode context from training data, they can predict the most likely next words in real time and offer suggestions tailored to a specific context. Tools built on these models can help people with aphasia more effectively, faster, and more accurately than existing methods.

The beneficiaries include people with aphasia, their caregivers, clinicians, and the broader healthcare system, with anticipated improvements in everyday communication outcomes. It also holds potential for people with dementia and older adults, where speech difficulties can also arise. Long-term benefits of fitting tooling include greater social participation, reduced isolation, and more efficient rehabilitation support.

What was your personal motivation to conduct this research?

Throughout my studies and work as a Teaching Assistant, I’ve always been interested in language and speech. Investigating the use of AI models for people with Aphasia allowed me to focus on real-world applications of AI and, more importantly, to contribute to a line of research that positively impacts society. A valuable lesson for me came from this explicit focus on impact, which taught me to judge success not only by performance metrics, but also by potential usefulness for people with aphasia and clinicians.

This shifted my perspective from “what is technically possible?” to “what is feasible, acceptable, and valuable in practice?”, a learning effect that I have applied many times since. Additionally, by working with supervisors from different faculties, I learned how to collaborate across disciplines, reconcile different expectations and vocabularies, and integrate technical, linguistic, and clinical perspectives into a coherent research plan.