By: H2O.ai Team
A drug discovery AI solution from H2O.ai Health
Powered by NVIDIA GPUs and NVIDIA AI
In a healthy individual, each cell type has its own metabolic program, carrying out specific functions. This organization is disrupted in disease, either as a cause or a result of it, or both, and this disruption is reflected in the patient’s gene expression profile.
Due to its high discriminatory ability, gene expression profiling is increasingly used as a biomarker for disease, disease stage, and drug compatibility. However, strongly non-linear relationships and patient-to-patient variations place a barrier to the accuracy with which statistical methods can infer a patient’s risk. Machine learning has been proven to be more sensitive and accurate in predicting patient risk from expression biomarker data.
Expression Biomarker AI is a machine learning-powered risk and classification application. Using gene expression information in samples derived from patient biopsies, the application allows the user to:
- Evaluate the patient risk and understand how common this risk has been among past patients (Driverless AI autoML).
- Handle very wide datasets (Lasso, PCA, Driverless AI).
- Find and explore similar patients (based on Euclidean distance).
- Identify the genes that influence patient risk (Shapley values).
- Explore potential interventions via model-based simulations (individual conditional expectation).
- Evaluate the consistency of the predictions among different demographic groups (disparate impact analysis).
The 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 is powered by the H2O AI Cloud: Driverless AI AutoML, H2O-3, MLOps, and H2O.ai Wave©, as well as by NVIDIA GPUs.