Skip to main content
Aurélien Vannieuwenhuyze

Expertise · Applied research · Evaluation

From clinical challenges to robust AI solutions

I support healthcare, research, and innovation teams in framing, designing, and evaluating artificial intelligence solutions where temporal dynamics, explainability, and domain knowledge are critical.

Based in Lille, France · Focused remote or hybrid engagements · International collaborations compatible with completing my PhD.

Areas of expertise

Specialized expertise for projects where time, knowledge, and explainability matter

My contribution sits between scientific research and engineering: clarifying the question, assessing feasibility, selecting a credible modeling strategy, and producing results that technical and domain teams can evaluate and use. Prototyping is introduced when the project maturity calls for it.

01

Scientific and technical framing

Problem formalization, data assessment, identification of clinical constraints, hypothesis selection, and definition of measurable success criteria.

02

Temporal clinical modeling

Medical time series, longitudinal events, patient trajectories, irregular data, and early-signal detection in complex clinical settings.

03

Explainable and neuro-symbolic AI

Combining machine learning, knowledge graphs, expert rules, and temporal reasoning to produce interpretable, explainable, and reviewable results.

04

Prototyping and evaluation

Proof-of-concept design, experimental protocols, model comparison, error analysis, and explicit documentation of limitations before industrialization decisions.

Engagement formats

Engagements adapted to the project’s level of maturity

An engagement may be short and exploratory or part of a longer applied R&D collaboration. Scope is defined around the decision the project needs to support.

  1. 01

    Diagnostic and framing

    Clarify the need, assess feasibility, identify risks, and build a scientific and technical roadmap.

  2. 02

    Feasibility study or proof of concept

    Test a hypothesis on a limited scope with an explicit protocol, metrics, and stopping criteria.

  3. 03

    Applied research collaboration

    Explore a methodological question, design an experiment, and turn results into a demonstrator, report, or publication.

  4. 04

    Part-time R&D support

    Bring specialized expertise into an existing team without requiring a full-time hire while I complete my PhD.

Deliverables

Results designed to support a decision

Each deliverable should clarify what works, what remains uncertain, and which decision should follow. Its form depends on project maturity: framing, feasibility study, experiment, prototype, or transfer.

  • Framing note and formulation of the scientific or clinical question.
  • Mapping of data, constraints, biases, and conditions for reuse.
  • Experimental protocol, metrics, and validation criteria.
  • Reproducible prototype or documented proof of concept.
  • Analysis of performance, errors, uncertainty, and limitations.
  • Technical recommendations and transfer to the internal team.

Who I work with

Collaborations with teams connecting data, medicine, and innovation

MedTech and digital health

AI feature framing, proofs of concept, patient trajectories, and explainability for healthcare products.

Clinical and hospital teams

Longitudinal data exploration, needs formalization, and applied research design.

Pharma, CROs, and medical devices

Temporal dynamics, signal evaluation, and methodological support for healthcare data projects.

Research labs and R&D consortia

Methodological collaboration, demonstrators, publications, and transfer between academic research and industry applications.

Working method

A short path from the question to the experiment

The work progresses through explicit stages to avoid premature development and preserve scientific traceability.

01

Understand

Context, users, target decision, available data, and domain constraints.

02

Frame

Hypotheses, scope, metrics, risks, and conditions for success or stopping.

03

Experiment

Baseline, prototype, comparisons, error analysis, and robustness testing.

04

Transfer

Documentation, findings, recommendations, and the next technical decisions.

Responsible approach

In healthcare, a convincing R&D approach must also reveal its limitations

I favor work where assumptions, data, metrics, and uncertainties remain explicit. A proof of concept is neither a validated medical device nor a substitute for clinical judgment.

  • Traceability of data, transformations, experiments, and decisions.
  • Evaluation beyond a single score: errors, subgroups, stability, and domain relevance.
  • A clear separation between R&D exploration, clinical validation, and regulated industrialization.

Professional credentials

Experience

Nearly 20 years across technology, data, and digital transformation

Research

PhD candidate in artificial intelligence at CNAM, CEDRIC Laboratory — ISID team

Clinical knowledge

University Diploma in Critical Care, University of Lille - awarded in 2026

International visibility

International publication at ER 2025 - Springer

Start a conversation

Do you have a clinical challenge, data, or an AI project to structure?

Share the context, the decision you need to support, and the data currently available. I can quickly indicate whether my expertise fits the problem and what a sensible first scope could be.