Blog

Blog

Reinventing How We Work with Medication Instructions

Healthcare professionals should be able to write naturally. Posologie explores how a small, specialised AI model can turn medication instructions into structured data and reduce repetitive form filling.

Clinical AINatural language processingHealthcare workflows
Read post

Why I Built MedDeID—and Why It Should Become a Shared Effort

Clinical data access became a bottleneck in my own PhD. MedDeID is my attempt to turn that bottleneck into open infrastructure for the wider healthcare ecosystem.

De-identificationClinical AIOpen source
Read post

From Clinical Text to Reusable Data: Reflections from the BIHR Summer Day

What two clinical text structuring projects taught me about the path towards a useful Belgian Integrated Health Record.

BIHRAI in HealthClinical NLP
Read post

Teaching AI to Future Physicians

A guest lecture on AI in medicine for final-year medical students in the University of Antwerp 'Terugkomdagen' program.

AI in HealthTeaching
Read post

Building a Production-Grade De-identification Annotation Server

What I learned while turning a clinical text de-identification model into a scalable API with GPU inference, batching, deployment automation, and Kubernetes scaling.

De-identificationClinical AIInfrastructure
Read post

Four Requirements for AI in Health

AI in healthcare only works when data, compute, clinical expertise, and governance are all in place together.

AI in HealthHigh-level thinking
Read post