Incoming Postdoctoral Researcher in AI for Robust Clinical Prediction Modelling, Big Data Institute, University of Oxford
PhD, ETH Zürich · Sensory-Motor Systems Lab & SCAI Lab
I build the parts of an explainable clinical digital twin: models of individual patients that keep their state in a form a clinician can inspect, predict how it changes, and answer what-if questions about treatment. My work joins two causal traditions that are usually kept apart — structure learning with graphical models, and potential-outcomes reasoning — with deep temporal models that are interpretable by design rather than explained after the fact.
I completed my doctorate at ETH Zürich with Prof. Robert Riener, Prof. Julia Vogt, and Dr. Diego Paez-Granados, working closely with clinicians at the Swiss Paraplegic Centre. I will join Prof. Christopher Yau's group at the University of Oxford to work on simulation-based pre-training for robust and trustworthy clinical prediction models.
I successfully defended my doctoral thesis, Towards Explainable Digital Twins for Clinical Decision Support across Spinal Cord Injury and Intensive Care, at ETH Zürich on 10 September 2026.
Jul 2026
I will join the University of Oxford as a Postdoctoral Researcher in AI for Robust Clinical Prediction Modelling, working with Prof. Christopher Yau at the Big Data Institute.
Jan 2026
Two co-first-author papers accepted at ICLR 2026: GARLIC, on interpretable graph attention for intensive-care time series, and BioBO, on biology-informed Bayesian optimization for perturbation design.
Dec 2025
KarmaTS was presented at ML4H 2025 in San Diego, supported by an ML4H Travel Award.
Oct 2025
Oral presentation at the ISCoS Annual Meeting in Gothenburg on graphical modelling for pressure-injury risk prediction.
Sep 2025
Completed a PhD research internship in Data Science & AI at Johnson & Johnson Innovative Medicine (June–September 2025), where BioBO was developed.
Apr 2025
Our team placed first of 14 in the ASIA Data Science Challenge on predicting functional recovery after traumatic spinal cord injury.
What I work on
Research
A clinical record tells you what happened — never what would have happened. My research builds models that clinicians can question anyway: graphs that say which factors carry risk, temporal models whose explanation is the model itself, and simulation frameworks that manufacture the ground truth needed to check a causal claim.
Medicine will not be improved by models that take the decision away from the clinician. It will be improved by models a clinician can work with — and nobody can work with a model they cannot question.
— from my doctoral thesis (ETH Zürich, 2026)
Causal structure learning for real clinical data
Rehabilitation cohorts are small and mixed-type, with gaps that mean something and causes that were never recorded. I develop structure-learning methods that admit latent confounders and mixed variables — including a predictive permutation conditional independence test — and use the learned graph as both explanation and feature selector. Applied to hospital-acquired pressure-injury risk in spinal cord injury, this reached a nested AUROC of 0.952; causal fusion then lets small, mismatched cohorts inform one another about structure rather than about patients.
Post-hoc explanations of clinical time-series models are unstable and often disagree with one another. GARLIC predicts intensive-care outcomes from irregular, incomplete trajectories, and its own attention weights and learned graph edges are the explanation — passing faithfulness tests that the interpretable baselines fail, while outperforming the baselines tested on three ICU benchmarks.
GARLIC · ICLR 2026
Simulation and known-truth evaluation
No architecture can validate a counterfactual claim against a record that never contains one. KarmaTS makes discrete-time causal processes executable, so temporal methods can be graded against dynamics that are known because someone wrote them down. ICareDT writes matched counterfactuals onto real intensive-care histories, letting treatment-effect estimates be scored against a known answer — cutting error by more than an order of magnitude versus the naive factual contrast. This line continues at Oxford in simulation-based pre-training for robust prediction models.
KarmaTS · ML4H 2025ICareDT
Decision-making under uncertainty in the laboratory
The same causal ideas steer experiments. BioBO brings biology-informed priors — from gene embeddings and enrichment analysis — into Bayesian optimization for perturbation design, letting prior beliefs guide exploration while their influence fades as evidence arrives, under a bound that limits what a wrong belief can cost. Developed during my internship at Johnson & Johnson Innovative Medicine and evaluated on the GeneDisco CRISPR benchmarks.
BioBO · ICLR 2026
Papers
Publications
Published & accepted
GARLIC: Graph Attention-based Relational Learning of Multivariate Time Series in Intensive CareICLR 2026Y. Li*, R. Wang*, M. Günther, D. Paez-Granados
BioBO: Biology-Informed Bayesian Optimization for Perturbation DesignICLR 2026Y. Li*, T. Cui*, T. Mansi, M. Prakash, R. Liao
The ASIA Data Science Challenge — Predicting Functional and Neurological Recovery from Acute ISNCSCI ScoresTSCIR 2026J. Villines*, R. Stirnimann*, L.-P. Lukas, O. Taran, M. Tuci, Y. Li, C. R. Jutzeler, J. L. K. Kramer, F. Geisler, D. Bourbeau, R. J. Cotton†, S. C. Brüningk†, and the ASIA Engineering and Data Science Committee
KarmaTS: A Universal Simulation Platform for Multivariate Time Series with Functional Causal DynamicsML4H 2025Y. Li*, H. Li*, D. Paez-Granados
Automated Prediction of Item-Level ARAT Scores from Wearable SensorsICORR 2025T. Weikert, Y. Li, D. Paez-Granados, C. Awai Easthope
Mixed-Variable Graphical Modelling Towards Risk Prediction of Hospital-Acquired Pressure Injury in Spinal Cord Injury IndividualsScientific Reports 2024Y. Li, A. Scheel-Sailer, R. Riener, D. Paez-Granados
Robust Feature Selection for Continuous BP Estimation in Multiple Populations: Towards Cuffless Ambulatory BP MonitoringIEEE JBHI 2024A. Cisnal, Y. Li, B. Fuchs, M. Ejtehadi, R. Riener, D. Paez-Granados
Causal Domain Adaptation via Perturbed DAG ReconstructionarXiv 2022Y. Li, T. Hatt, I. Bica, M. van der Schaar (MSc thesis, ETH Zürich)
A Closer Look at the Training Strategy for Modern Meta-LearningNeurIPS 2020J. Chen, X. Wu, Y. Li, Q. Li, L. Zhan, F. Chung
Improvement of Embedding Channel-Wise Activation in Soft-Attention Neural Image CaptioningICVISP 2018Y. Li
Under review
ICareDT: Towards Personalized Causal Digital Twin for Intensive Carenpj Digital MedicineY. Li, Y. Qi, D. Paez-Granados
Anaerobic Microbial Communities Influence Wound Recovery Outcome in Pressure Injuries: A Machine Learning ApproachScientific ReportsY. Li, J. Stoyanov, R. Riener, R. Wettstein, S. Capossela, D. Paez-Granados†, A. Bertolo†
Patient Digital Twins for Caregiving RobotsBook chapterD. Paez-Granados, Y. Li, M. Ejtehadi, B. Fuchs, O. Gnarra (Springer Moonshot Consortium)
In preparation
Causal Graphical Fusion with Attention Steering for Incomplete Spinal Cord Injury DataY. Li, D. Al Jorf, A. Ighissou, A. Scheel-Sailer, R. Riener, D. Paez-Granados
Collaborative Causal Discovery with Multimodal Data in Spinal Cord InjuryY. Li*, S. Zhong*, H. Li, R. Riener, D. Paez-Granados
GRITS: Graphical Imputation for Multivariate Time SeriesS. Joray, Y. Li, D. Paez-Granados
Monocular Markerless Biomechanics for Clinically Interpretable Gait Assessment in Spinal Cord InjuryS. Natraj, M. Ruepp, Y. Li, R. Riener, D. Paez-Granados
* equal contribution · † shared senior authorship
Presentations
Talks & posters
Apr 2026ICLR 2026 — GARLIC and BioBO (posters)
Rio de Janeiro, Brazil
Dec 2025Machine Learning for Health (ML4H) — KarmaTS (poster)
San Diego, United States
Oct 2025ISCoS Annual Meeting — Mixed-variable graphical modelling towards risk prediction of hospital-acquired pressure injury (oral)
Gothenburg, Sweden
May 2024AI for Good Global Summit — A digital twin framework for chronic diseases (booth)
Geneva, Switzerland
Apr 2023European Causal Inference Meeting — Continual causal discovery and inference for SCI comorbidities (poster)
Oslo, Norway
Academic service
Teaching, mentoring & review
Teaching
Artificial Intelligence for Healthcare and Rehabilitation, ETH Zürich Teaching Assistant — Autumn 2024 & Autumn 2025
Peer review
Conferences: NeurIPS, ICLR, ICML, AISTATS, ML4H
Journals: Journal of NeuroEngineering and Rehabilitation
Student supervision
2026D. Al Jorf — Master's lab practice, ETH Zürich
2025R. Wang — Master's thesis, University of Zurich
2025Y. Qi — Master's thesis, ETH Zürich
2025S. Joray — Master's internship, ETH Zürich
2024A. Ighissou — Master's thesis, University of Milano-Bicocca
2024J. Koch — Semester project, ETH Zürich
2023H. Li — Master's internship, TU Munich
2023B. Wu — Master's thesis, TU Munich
Recognition
Awards & scholarships
Nov 2025
✦Machine Learning for Health (ML4H) Travel Award
Apr 2025
✦First place (of 14 teams), ASIA Data Science Challenge — functional recovery prediction after traumatic spinal cord injury
Feb 2018
✦Outstanding Student Award, Faculty of Applied Science & Textiles, PolyU — top graduate of the faculty; BSc GPA 4.00/4.00
2014–18
✦Dean's Honours List (top 5%), Faculty of Applied Science & Textiles, PolyU · Ho & Ho Foundation Entry Scholarship for outstanding students
May 2017
✦Bauhinia Cup Outstanding Entrepreneurs Association Scholarship
Oct 2016
✦Commercial Radio 50th Anniversary Scholarship
Jun 2015
✦Reaching Out Award, Hong Kong Government Scholarship
