Topic area: Case Study
Current bio-informatics tools do not capitalize on the great advancements made in Machine Learning (ML) that can enable them to generate more, and more rapid, breakthroughs. Big Data technologies can facilitate the complete integration of heterogeneous sets of experimental data to identify key metabolic pathways and drug targets to enable precision medicine.
The Holy Grail in precision medicine is the integration of multi-omics and clinical phenotype data within clinical investigations to drive tailored therapies for patients. Artificial Intelligence (AI), and specific applications such as Deep Learning are successfully deployed in a variety of domains, and show similar promise when applied to the data challenges of precision medicine. Standard machine learning methods, however, were built solely on data-driven approaches where the number of samples greatly exceeds the number of dimensions. In biomedical research, these techniques need to be adapted to be effective on the transpose problem (ultra high-dimensionality, very low number of samples). In this talk, we will explore some avenues for these adaptations and present use cases of them in action on data from the National Institutes of Health.