AI for Clinical Decision Support and Health Services

My work in clinical decision support studies how information from patient records and clinical guidelines can inform a clinician’s decision process. The HyCARE research direction combines clinical information extraction with guideline-based decision support, with uncertainty in model outputs examined through evaluation.

I also develop models for health services, including referral networks and emergency-department demand. This work uses network models and time-series forecasting to study how patients move through services and how demand changes over time. C-PATH studies conversational assistance for patient triage.

Causal explanations for graph neural networks provide methods for examining model predictions. Work on medical vision-language representations evaluates image-text retrieval. RELIABLE extends the application scope to population health risk during extreme weather events.

Funded projects

Projects active during 2021 or later, including completed projects.

• HyCARE (2025–2026). Principal investigator. Clinical information extraction and decision support combining clinical guidelines with machine learning, with use cases in radiology scheduling and lung cancer treatment. Reference 2024.07361.IACDC.
• TRUST-DL (2022–2023). Principal investigator. Causal explanations for specialist referral recommendation. Reference FCT/CPCA-IAC/2022/01.
• MD2TRUST (2022–2023). Principal investigator. Trustworthy data science for healthcare efficiency, focused on specialist referral recommendation. Reference CMU/TIC/0016/2021.
• RELIABLE (2020–2022). Team member. Prediction of building occupants’ health risk during extreme weather events, combining health, weather and building data. Reference DSAIPA/DS/0111/2019.
• IPPU — Identifying and Predicting Emergency Admissions (2019–2023). Team leader. Forecasting emergency admissions to support health-service planning. Reference DSAIPA/AI/0087/2018.

Publications since 2021

• Dissecting medical referrals in health services: role of physician professional networks. BMC Health Services Research, 2026.
• Hierarchical Spatio-Temporal Transformer for Coherent Emergency Department Forecasting. Accepted at SoGood, ECML PKDD 2026.
• C-PATH: Conversational Patient Assistance and Triage in Healthcare System. 2025 IEEE International Conference on Digital Health (ICDH).
• Causal Inference Explanations for Graph Neural Networks. Causal@UAI2024 Poster.
• A Temporal Fusion Transformer for Long-Term Explainable Prediction of Emergency Department Overcrowding. Machine Learning and Principles and Practice of Knowledge Discovery in Databases, 2023.
• Referral prediction in Healthcare using Graph Neural Networks. GReS – Workshop on Graph Neural Networks for Recommendation and Search. Co-located with the ACM RecSys ’21 conference, 2021.

Preprints and related manuscripts since 2021

• CIExplainer++: Generating Causal and Interpretable Explanations for Graph Neural Networks. arXiv preprint, 2026.
• Beyond Symmetric Alignment: Spectral Diagnostics of Modality Imbalance in Vision-Language Models in the Medical Domain. arXiv preprint, 2026.
• Can Large Language Models Reliably Extract Physiology Index Values from Coronary Angiography Reports?. arXiv preprint, 2026.
• Extreme Multilabel Classification for Specialist Doctor Recommendation with Implicit Feedback and Limited Patient Metadata. arXiv preprint, 2023.