Scott Wood

The paper develops a multi-omic random forest model to predict immunotherapy response in melanoma using SHAP for feature threshold identification.

This paper integrates tumor homologous recombination deficiency status with germline BRCA1 and BRCA2 variant classification to improve pathogenicity prediction.

This paper benchmarks Bayesian deep learning models for predicting cancer of unknown primary, showing improved uncertainty generalisation and accuracy under distributional shifts.