Carlos Vega

Carlos Vega

PhD · Research Engineer in Digital Health

Luxembourg Institute of Health


Data science, digital health, philosophy of science & data ethics.

I also teach Applied Philosophy of Science & Data Ethics at the University of Luxembourg.

Learn more

Now

I work as a Research Engineer in Digital Health at the Luxembourg Institute of Health, developing data integration, software and data architectures for clinical studies and biomedical research.

Since 2021, I have directed and taught Applied Philosophy of Science & Data Ethics in the Master in Data Science at the University of Luxembourg.

As an MIT Catalyst Europe fellow, I co-lead two health-innovation projects: improving emergency response for cardiac-device patients and tackling period-poverty-related school absenteeism in Ghana. Each project has secured ≈ €50,000 in funding.

Experience

  1. Luxembourg Institute of HealthResearch Engineer in Digital Health
  2. Luxembourg Centre for Systems BiomedicinePostdoctoral Researcher
  3. UAM & Naudit HPCNR&D Computer Science Engineer
  4. Universidad Autónoma de Madrid (UAM)Scholarship Assistant

Selected work

Things I've built and published, grouped by theme.

Scientific reasoning & trustworthy AI

Philosophy of science helps me address concrete problems in data-driven medicine. I published a framework for instructors that connects induction, causation, explanation, bias, confounding and ethics to data-science practice. From Hume to Wuhan explains how induction and missing domain knowledge can undermine the transfer of medical ML models to new hospitals or clinical settings. The Monkeypox dataset analysis showed how published classifiers learned dataset artefacts rather than disease-related signals.

Biomedical knowledge & translational ML

BioKC links literature evidence with structured systems-biology knowledge for collaborative curation. My work on causal metadata shows how causal assumptions and confounding can be represented explicitly in biological datasets. Related work on translational ML assesses why models developed in research settings can fail when moved into clinical or laboratory use.

Privacy-aware & frugal AI

CellAnalyzer runs cell-image models locally in the browser with ONNX, so images never leave the user's machine. The Leukemia Registry Assistant uses client-side embedding models to assist with registry form filling, so sensitive clinical data never leave the user's machine.

Clinical research software

SMASCH coordinates longitudinal-study visits across participants, staff, assessments and resources. It has been in use since 2018 in programmes including NCER-PD, PDP, RBD, MCI-biome and CON-VINCE.

Earlier systems engineering

During my PhD, I worked on high-performance network-data processing and anomaly detection. Examples include HTTPanalyzer, which processes traffic above 10 Gbps on commodity hardware, and KISS Methodologies for Network Anomaly Detection.

More research

Recent publications also cover agentic-AI safety in mental health and reproductive health, alongside other clinical and biomedical research.

Education

Computer Science Engineering — Universidad Autónoma de Madrid
BSc 2013 · MSc 2014 · PhD 2018 · PhD cum laude with industrial mention

My doctoral research focused on high-performance computing for large-scale data collection, processing, analysis and visualisation, including multi-Gbps network traffic and anomaly detection in collaboration with industry.

Further study: Philosophy of Science — University of Oxford, Department for Continuing Education.