Causal Effects of Endogenous and Exogenous Risk Factors for Wasting

Scalable online super-learner for predicting health outcome phenotypes and an online scalable super learner of an optimal individualized treatment rule for improving these health outcomes.

CTML members involved: Mark van der Laan Ph.D., Alan Hubbard Ph.D., Jeremy Coyle Ph.D. Andrew Mertens Ph.D., Ivana Malenica

The overall goal of this grant would be to develop a scalable online super-learner for predicting health outcome phenotypes and an online scalable super learner of an optimal individualized treatment rule for improving these health outcomes. These will represent tools for the Gates Foundation to optimize the delivery of the right intervention, to the right child, at the right time and right price.

Led by Mark J. van der Laan and Alan E. Hubbard, this research approaches using weighted combination of many candidate learners to build super learner predictive algorithms. Along with the Policy Delivery and Implementation Surge Team  (PDIST), this team will build a platform for consuming models from other surge teams, longitudinal and cross-sectional child growth and development data. The Policy Delivery and Implementation Surge Team (PDIST) aim is to analyze growth and development outcomes across geographic, regional, cultural, and socio-economic secular trends that contribute to poor growth outcomes in order to support our ability to promote healthy birth, growth, and development in the communities that need it most by delivering the right intervention(s), to the right child, at the right time, and at the right price.