Research summary
By using machine learning approaches to integrate multi-omic data with diverse phenotypes from healthcare data, researchers may be able to better understand complex human diseases. (Multi-omics combines the datasets of different “omic” groups—such as genome, proteome, transcriptome, epigenome and microbiome—during analysis.) But several key challenges hinder this work, including concerns about patient data privacy and challenges related to scaling and interpreting models to support advances in precision medicine.
Dr. Yue Li, Canada Research Chair in Machine Learning for Genomics and Healthcare, is working on machine learning developments to support federated learning with electronic health records. (Federated learning is a way to train artificial intelligence models without seeing or touching the data.) He and his research team are also looking at single-cell multi-omic modelling and neurodegenerative diseases and studying the genetic variants involved in complex diseases. They hope to make significant contributions to the field of computational medicine.