Nura Brimo

This paper introduces a hierarchical temporal transformer for cancer grade prediction from pathology reports, showing improved performance across cancer types and using visit timing when identifiable.

This paper evaluates optimiser repairs in living-cell self-driving labs, showing when they improve performance and when constraints make them redundant.

The paper evaluates 25 electrospun PVP fiber formulations for regenerative endodontics, using machine learning to identify constraints and release risks, finding carrier-limited release and recommending further testing.

This paper compares measurement-aware models and selection-level repairs in Bayesian optimisation for living-cell experiments, finding that measurement-aware models reduce regret and improve coverage under data contamination.

The paper introduces HTT, a model that improves cancer grade prediction from pathology reports by incorporating temporal information and cross-cancer transfer learning, achieving high AUROC scores on unseen cancer types.

The paper uses machine learning to analyze a dataset of electrospun nanofibers, finding that medicament identity, not total loading, better predicts fiber morphology in regenerative endodontics.