Deep Learning–Based Estimator for the Non-Iterative Conditional Expectation (NICE) g-Formula
Join us on Wednesday, October 22 for the “Works In Progress” Epidemiology Seminar Series, featuring Dr. Jing Li discussing Deep Learning–Based Estimator for the Non-Iterative Conditional Expectation (NICE) g-Formula .
Abstract: The g-formula can be used to estimate causal effects of sustained treatment strategies using observational data under the identifying assumptions of consistency, positivity, and exchangeability. The non-iterative conditional expectation (NICE) estimator of the g-formula also requires correct estimation of the conditional distribution of the time-varying treatment, confounders, and outcome. Parametric models, which have been traditionally used for this purpose, are subject to model misspecification, which may result in biased causal estimates, particularly in high-dimensional or nonlinear settings. To address these limitations, we propose a unified deep learning framework for the NICE g-formula that leverages recurrent neural networks to flexibly model the joint conditional distribution of time-varying variables. Through simulation studies, we demonstrate the advantages of the proposed deep learning–based estimator over conventional parametric approaches and establish its statistical validity through uncertainty quantification.
Bio: Dr. Jing Li is a Postdoctoral Research Fellow at CAUSALab in the Department of Epidemiology at the Harvard T.H. Chan School of Public Health, working under the mentorship of Prof. Miguel Hernán. Her research focuses on combining causal inference and machine learning to better understand the effects of treatments and interventions in complex longitudinal settings. Her current work includes advancing g-formula methods, developing deep learning techniques to improve the estimation of sustained treatment effects, and building large-scale simulation experiments to evaluate the performance of deep learning-based estimators. Her work has appeared in leading venues such as CVPR, ICLR, AAAI, IEEE Transactions on Image Processing (TIP), and NeurIPS workshops. In addition to academic publications, she is the primary developer of the open-source Python package pygformula, and also contributes to several other causal inference software packages, including gfoRmula and gfoRmulaICE. Jing Li received her Ph.D. in Computer Science from Peking University, where she was advised by Prof. Yizhou Wang in the School of Computer Science. She received her bachelor’s degree in Statistics from the School of Mathematics and Statistics at Wuhan University.
Speaker Information
Jing Li, PhD
Organizers
Related Events
ⓘ Harvard Chan School hosts a diverse array of speakers, invited to share both scholarly research and personal perspectives. They do not speak for the School, and hosting them does not imply endorsement of their views, organizations, or employers.