The rapid advancement of genomics will make it feasible to enhance clinical trial design and success rates using patient-level pharmacogenomics reports for clinical trials participants.
In this research, Evaluation of Pharmacogenomic Markers of Drug Response Using Machine Learning Approaches, the research team from Georgetown University's Innovation Center for Biomedical Informatics, working in collaboration with the FDA Center for Drug Evaluation and Research, used existing or newly developed bioinformatics tools to integrate omics data from university-held databases and other sources.
This integrated view of the data allowed for the investigation of the effects of genomic variation on drug sensitivity, resistance, and mechanisms of action. Results from this analysis led to recommendations on usage of omics data in clinical decision support, drug dosing, adverse event detection and clinical trail design, including recommendations on what and when genomics information should be considered during drug development and regulatory review.
The research team also studied the association of variations in human whole genome and whole exome data sets with information about drug efficacy, toxicity, and adverse events. Genomic information on subjects during early drug development could allow for discovery of genomic differences – before the drug goes to market – to potentially improve the efficacy, safety, and prescribing/marketing practices for a drug. Currently, most data regarding adverse events, differences in effectiveness, and dosing are found after the drugs have been on the market.
"Machine learning" is a scientific discipline concerned with the design and development of algorithms to allow computers to evolve behaviors based on empirical data. A major focus of machine learning research is to automatically recognize complex patterns and make intelligent decisions based on existing data.