From the release:

Multi-scale deep learning and single-cell models of cardiovascular health

The team will develop methods to accelerate the pace of discovery of genetic determinants for cardiovascular disease. They will develop new statistical machine learning tools to analyze morphological and functional parameters of the heart from clinical images, an approach that can be scaled to analyze millions of images. They will also develop machine learning tools based on enhanced iterative random forests (iRF) to identify genetic variants likely to account for some of the variation in cardiovascular morphology and function observed in their analysis of clinical images, utilizing publicly available large-scale clinical data sets and local patient cohorts. Finally, they will identify genetic variants responsible for functional phenotypes using cell-based in vitro model systems.

Ben Brown and Bin Yu's team

Machine learning for interpreting rare genetic variation in comprehensive newborn screening and pharmacogenetics

In California, 500,000 babies are born each year, some of whom have genetic mutations that cause disease or altered responses to medications. Recognizing which genetic variants cause problems is surprisingly difficult, impeding the use of genetic information to inform early intervention or the customization of patient care. The team has drawn together experts in biology, computer science, medicine, and ethics to develop new methods for identifying genetic variants that cause disease, focusing on serious newborn diseases and on gene variants that affect patient responses to medications. The team will collect experimental data and develop innovative machine learning techniques to predict the functional consequences of genetic variants.

Mike Jordan's team