GenomeDx Biosciences is a clinical genomics company that uses machine learning on genomic data to develop clinical tests for cancer. They developed a genomic classifier to predict prostate tumor Gleason grade using RNA expression data from over 7,000 patients. This Gleason grade classifier was tested on a separate dataset and achieved an AUC of 0.77, outperforming other clinical predictors. The classifier also predicted metastatic outcomes with an AUC of 0.73, demonstrating its ability to predict patient risk. GenomeDx uses the H2O platform for its machine learning work due to its ability to handle high-dimensional genomic data and its deep learning algorithms, which can model complex nonlinear relationships between genes.