Clinical Machine Learning
Reproducible workflows for assisted-reproduction data, including outcome definition, feature construction and patient-aware, time-aware validation.
Real-world data in assisted reproduction
Since June 2026, I have worked as a Research Assistant at the University of Surrey, developing reproducible machine-learning workflows for large, heterogeneous clinical datasets in assisted reproduction and in vitro fertilisation (IVF). This research involves collaboration with Dr Xilu Wang at Surrey and Dr Mingkun Tan at the University of Münster, Germany.
This ongoing research focuses on making observational data suitable for reliable outcome modelling:
- Outcome definition: define clinically meaningful endpoints while accounting for incomplete records and differences in outcome availability.
- Feature construction: create analysis-ready features that respect the longitudinal structure of treatment cycles.
- Validation: design patient-aware and time-aware evaluation to reduce information leakage and assess model generalisation.
- Reproducibility: maintain auditable workflows with attention to data provenance and the confidential handling of de-identified healthcare data.
The work extends my research in uncertainty-aware modelling and temporal prediction to real-world clinical observations.