Sam is focussing on applying computational analysis to explore lymphoma medical data. This involves producing neural networks to expedite clinical workflows and identify features correlated with clinical labels.
Previously, Sam studied Physics at Durham University, specialising in Astronomy, dealing with large datasets, scientific computing and mathematics. Subsequently, he worked as a data scientist using tools such as R, SQL and specialist geographical software working in a close-knit team. After this, he studied at the University of St Andrews and was able to focus on both AI and medicine.
My main research areas of interest are:
- Developing explainable, state-of-the-art deep learning models.
- Applying A.I. on rich, complex medical data to assist clinicians.
Specifically, I am using neural networks and AI to process high resolution histological lymphoma data to predict classifications in an explainable and generalisable manner. Interpretability is important to both the researcher and the clinical end-user. One example is when training networks it is necessary to understand which datapoints the model fails to process as desired. Another example where interpretability is critical is when conducting unsupervised research, so we can better semantically understand any novel biomarkers identified.
To this end, I am developing multiple smaller models to focus on different aspects of the data, so then each model can clearly answer different semantic questions, such as how the spatial arrangement of cells affects the output, or whether there is clinically relevant information embedded in the cellular morphologies observed in the tissue.
- MSc Physics and Astronomy, Durham University, UK
- MSc Advanced Computer Science, University of St Andrews, UK