Research · Newcastle
Predictive modelling under real constraints.
Clinical prediction problems where the data is imperfect, the cohorts are small, and the consequences of being wrong are material.
My research applies machine learning to clinical prediction problems where the data is imperfect, the cohorts are small, and the consequences of being wrong are material. The current focus - my second-year PhD at Newcastle University - is a predictive framework for limbal stem cell deficiency, a corneal surface disease whose trajectory is hard to forecast from standard imaging alone.
Prior work spans clinical simulation (a chronic kidney disease ML model at HealthLumen, 2020) and Arabic-language health-NLP (a peer-reviewed paper in the Journal of Islamic Marketing, 2021).
I welcome co-authorship, clinical partnerships, and reviewer invitations in ophthalmic AI, cohort-based clinical ML, and Arabic-language health NLP.
Publication co-authors.
Co-authors on the peer-reviewed Arabic health–NLP paper - links go to their Google Scholar profiles.
Posture
Methods stack: PyTorch and MONAI for medical imaging; Python for analysis; HPC compute through Newcastle for training runs that exceed local hardware. The data lives inside a UK clinical-research framework with explicit governance; the code, where licensing allows, is released alongside each paper.
- 3degrees in the data-science / ML chain
- 1peer-reviewed publication (DOI-linked)
- 4conference and seminar presentations (2025–26)
- 2hr → 7mclinical simulation speedup (HealthLumen, 2020)
Reply within five working days.