Research Projects
I develop AI and data science methods for sustainable, climate-resilient socio-technical systems, with a
focus on energy, transportation, environmental monitoring and responsible AI.
AI for Energy Demand & Climate-Resilient Systems
This line of work focuses on forecasting national and regional energy demand under climate and socio-economic
variability, with an emphasis on interpretable and operationally useful models.
- Hybrid SARIMAX, Prophet–LSTM, Seq2Seq and Transformer-based models for electricity demand in the Netherlands.
- Integration of subseasonal climate signals and economic indicators to improve wintertime energy planning.
- Uncertainty quantification and explainability (e.g. CRPS/Brier, SHAP, attention) to support transparent decision-making for grid operators and policymakers.
Transportation Analytics & Rail Operations
In transportation, I work on forecasting travel demand and service reliability in complex rail networks,
combining graph-based and temporal models with environmental data.
- Spatiotemporal models for delay and cancellation prediction in the Dutch railway network using multi-year operational and weather data.
- Graph neural networks and graph–sequence hybrids (e.g. GAT–LSTM, DGNN–LSTM) for network-aware predictions.
- Explainable AI for operations: attention-based explanations and SHAP analysis to help operators understand drivers of disruption.
- Leaf-fall and asset-condition modelling from satellite time series to anticipate infrastructure-related disruptions.
Environmental Monitoring & Groundwater Quality
This research applies time-series and transfer learning methods to environmental sensing data, with a focus
on sustainable water management.
- AI-based monitoring and quality prediction of groundwater in Zuid-Holland using LSTM and transfer learning.
- Fusion of hydrological, meteorological and operational measurements for robust water quality forecasts.
- Decision-support tools to assist regional authorities in managing environmental risks and monitoring regimes.
Fairness & Explainability in Medical Imaging
In healthcare, my work explores fairness-aware and explainable AI for clinical decision-support, especially
in medical imaging.
- Fairness-aware CNN and Vision Transformer models for chest X-ray analysis.
- Design and evaluation of fairness metrics and subgroup analyses to detect and reduce performance disparities.
- Uncertainty-aware and explainable predictions (saliency, attention, SHAP) to support trustworthy deployment in clinical workflows.
Interactive, Ubiquitous & Responsible AI Systems
Beyond domain-specific projects, I collaborate with industry and public partners on interactive and
infrastructure-oriented AI systems.
- MEC/IoT and human-in-the-loop ML for edge and ubiquitous computing, building on earlier work in green LTE and wireless sensor networks.
- Decision-support and forecasting systems with Dutch industry partners (e.g. pricing, operations and knowledge graphs in mobility and AEC sectors).
- Development of reproducible ML pipelines, data engineering workflows and dashboards for operational use.