Skip to main content
Aurélien Vannieuwenhuyze

Applied research · Experimentation · Transfer

Making my research visible, testable, and open to discussion

These demonstrators turn research questions into concrete experiments: interfaces, models, evaluation protocols, and explicit analyses of limitations.

Design principles

More than a technical showcase

Each project will be documented as an R&D experiment: problem, hypotheses, data, method, results, errors, limitations, and possible transfer paths.

01

Understandable

The problem and its clinical or industrial value should be clear without unnecessary jargon.

02

Reproducible

Choices, metrics, and experimental conditions should be open to examination.

03

Responsible

Limitations, uncertainty, and risks of misinterpretation are explicitly presented.

Roadmap

Explore the demonstrators

Demo 001 is available now. The other tracks remain presented with their actual level of progress.

01 Available

Demo 001 — Temporal Clinical Intelligence

An interactive demonstrator showing how a synthetic clinical trajectory can be structured as temporal events, enriched with OWL-Time, and explored as a graph to reveal relations, delays, and sequences that are not directly present in the source data.

Temporal knowledge graphOWL-TimeSynthetic clinical data
02 Planned

Explainable temporal prediction

A demonstrator for studying not only the probability of a clinical event, but also its time horizon, the signals contributing to the prediction, and the associated uncertainty.

Event predictionTime horizonUncertainty
03 Exploration

JEPA representations for medical trajectories

An exploration of latent-space predictive approaches for learning structured representations of clinical trajectories without reducing a patient to a static snapshot.

JEPASelf-supervised learningRepresentations
04 Exploration

Temporal neuro-symbolic reasoning

An experiment combining statistical learning and explicit knowledge to reason over clinical events, intervals, and temporal constraints.

Temporal graphsKnowledgeNeuro-symbolic AI

Publication framework

What will accompany each demonstrator

  • A description of the problem and intended audience.
  • The data used, preferably synthetic, public, or appropriately de-identified.
  • The method, technical stack, and architecture choices.
  • Metrics, error analysis, and explicit limitations.
  • A video, screenshots, or a testable interface whenever possible.
  • A public code repository only when intellectual-property and confidentiality constraints allow it.

Data and responsibility

No identifiable health data will be exposed

These projects are applied-research experiments. They are not validated medical devices, clinical recommendations, or evidence of effectiveness in real-world care.

R&D collaboration

Do you have a use case worth exploring?

A MedTech company, clinical team, or research laboratory can suggest a problem, dataset, or hypothesis to frame. The first discussion is used to assess feasibility and define a relevant scope.