Robert A. Dunne

This study identifies novel Alzheimer's disease genes and epistasis using machine learning GWAS tools, capturing more phenotypic variance and discovering new loci with epistatic interactions.

RFlocalfdr thresholds variable importance from Random Forests using an empirical Bayes approach, achieving high accuracy and speed for high-dimensional datasets.

This paper introduces a computationally efficient Random Forests method to detect epistatic interactions in genomic data and validates it on simulated and real datasets.

This paper introduces CursedForest, a Random Forest implementation that scales to millions of features and outperforms existing methods in speed.