Dr. David McCoy, a postdoctoral researcher at CTML, holds a Ph.D. in Environmental Health Sciences from the UC Berkeley School of Public Health, where he was mentored by Alan Hubbard, Alejandro Schuler and Mark van der Laan. His educational background also includes an MSc in Epidemiology, a BA in Philosophy, and a BS in Cognitive Neuroscience. David’s research is rooted at the intersection of causal inference, machine learning, and semiparametric statistical theory. With this multidisciplinary approach, he crafts robust solutions to complex, real-world problems. He has developed innovative methods, such as data-adaptive target parameters for mixed exposure analysis and mediation, that help researchers answer causal questions with statistical confidence. As the creator and developer of the CVtreeMLE and SuperNOVA packages, David employs cross-validated targeted maximum likelihood estimation in conjunction with data-adaptive target parameters, producing efficient estimators for mixed exposure and mediation analysis. A dedicated educator, David has taught Targeted Learning (Public Health 243A) and Advanced Topics in Causal Inference (PH252E) at UC Berkeley. Passionate about making intricate semiparametric theory accessible, he aims to engage a wide audience, including epidemiologists and ecologists. One of David’s overarching goals is to elevate the role of data-adaptive target parameters in research methodologies, particularly when key aspects of data—such as interacting exposures or modifying covariates—are unknown a priori. David’s work also extends into the realm of green chemistry, where he is developing interpretable machine learning models to guide toxicologists. These models aim to identify and avoid molecular structures that could bind to receptors in the human body. His recent semiparametric research includes leveraging the highly adaptive lasso for joint exposure estimation and enhanced interpretability, illustrating David's commitment to advancing both the science and the application of modern statistical methods.
Targeted Learning, Mixed Exposures, Mediation, Causal Inference, Machine Learning
