Objective
As precision medicine becomes more widespread, both medical diagnosis and drug discovery are increasingly relying on and leveraging the individual’s genomic and phenotypic profiles. H2O.ai’s Gene Mutation AI offers a portable personalized medicine interface that can offer an individual’s genetic risk on demand, present a selection of the ten most influential genes or pathways for this sample among thousands of genes, not dozens and play what-if scenarios to simulate dose-response outcomes for the selected patient.
Outcome
H2O.ai’s Gene Mutation AI application supports multiple pretrained models, so the same patient can be examined for their risk for multiple diseases or response to alternative pharmaceutical interventions. Gene Mutation AI can be easily extended to analyze multiple cancers, biopsy results, and molecule/drug responses. The app allows the user to load models trained with their own data, allowing detection of more types of malignant tumor.
Business Value
The solution helps to evaluate the risk of malignancy and understand how common this risk has been among past patients. It aids in identifying the cellular pathways that influence cancer malignancy risk and compare a patient’s predicted malignancy risk to that of previously evaluated patients.
H2O's AI and Data Approaches
Gene Mutation AI is powered by the H2O AI Cloud, Driverless AI AutoML, H2O-3, MLOps, and H2O.ai Wave, as well as by NVIDIA GPUs. The data science approaches include genetic algorithm, gene ontology-based feature engineering and advanced feature engineering, GLM, GBM, XGBoost, ensemble stacking, isotonic regression, Shapley value estimation, individual conditional expectation, and similarity search algorithms.