Razieh Nabi
Emory University, GA·Mathematics & Statistics
Data can show that two things move together, but not whether one causes the other. Knowing cause and effect is what lets doctors, regulators and engineers predict what will happen if they change something.
- Fields
Mathematics & Statistics
Public Health & Epidemiology
Computer Science
- Topics
causal inference
causal discovery
unmeasured confounding
- Grants
- 2 known grants, through Sep 2028
What they work on
- Identify causal effects as functions of observed data and estimate them robustly when unmeasured confounders — hidden variables that influence both the cause and the outcome — are present.
- Develop scalable, optimization-based (differentiable) methods for learning causal graphs, diagrams of which variables drive which, from noisy or partially observed data.
- Build a framework in which large language models extract insights from scientific literature and databases to validate assumptions or suggest modifications to causal model structure when observational data are sparse.
- Create dimension reduction techniques grounded in causal structural models that compress many correlated pollutants into summaries while preserving their causal link to health outcomes, with minimal distributional assumptions.
- Estimate short-term health effects of air pollution mixtures — criteria pollutants, fine particulate species and volatile organic compounds — using two decades of emergency department visit records from Atlanta.
More about the research
The question gets much harder when data are large, noisy or incomplete, or when important variables were never measured at all. Work here develops statistical and machine learning tools for drawing causal conclusions under exactly those conditions, and applies them to a concrete public health problem: how mixtures of air pollutants affect health.
Why it matters
Sharper estimates of how real-world pollution mixtures affect emergency department visits could inform air quality standards, risk assessment and multi-pollutant regulation, benefiting residents of cities like Atlanta. More broadly, causal methods that tolerate hidden variables and missing data could support more reliable decisions in healthcare, economics, engineering and public policy, where acting on correlation alone can mislead.
Known grants
From NIH and NSF research grants. Labs often have other funding we can't see.
Collaborative Research: MFAI: Two Sides of a Tapestry: Causal Inference and Model Discovery Amid Information Gaps in Complex Data
NSF · Oct 2025 – Sep 2028
Causally-Sufficient Dimensionality Reduction Methods for Assessing Joint Effects of Air Pollution Mixtures on Health Outcomes
NIH · Sep 2024 – Aug 2026
Closest labs
Faramarz Fekri
Georgia Institute of Technology, GA·Computer Science
Identify when cause-and-effect relationships can be calculated from observed data even when unmeasured confounders (hidden variables that affect both cause and outcome) are present, and develop robust ways to estimate those effects.
Leonard Schulman
California Institute of Technology, CA·Computer Science
Relax the statistical separation assumption needed when a hidden global confounder takes a limited number of values, aiming instead for model identification measured in Wasserstein distance, a way of comparing probability distributions.
Yang Ni
University of Texas at Austin, TX·Mathematics & Statistics
Develop a Bayesian joint model of treatment, outcome, covariates, and hidden confounders that identifies causal effects from observational data without instrumental or proxy variables.
The school
Emory University: 70 labs on PI Pages, $1.2B a year on research, 286 research doctorates a year.
Reach out
A start for your email. Fill in the brackets.