Machine Learning for Integrating Data in Biology and Medicine: Principles, Practice, and Opportunities

摘要: New technologies have enabled the investigation of biology and human health at an unprecedented scale in multiple dimensions. These dimensions include a myriad properties describing genome, epigenome, transcriptome, microbiome, phenotype, lifestyle. No single data type, however, can capture complexity all factors relevant to understanding phenomenon such as disease. Integrative methods that combine from thus emerged critical statistical computational approaches. The key challenge developing approaches is identification effective models provide comprehensive systems view. An ideal method answer biological or medical question, identifying important features predicting outcomes, by harnessing heterogeneous across several variation. In this Review, we describe principles integration discuss current available implementations. We examples successful medicine. Finally, challenges biomedical integrative our perspective on future development field.