Overview
I am a PhD student in Statistics at Texas A&M studying statistical inference when the underlying state is not directly observed and measurements are imperfect. My research focuses on identifiability, Bayesian hierarchical and latent-variable models, simulation-based validation, and uncertainty propagation.
Featured publication: Bayesian latent-class modeling for disease prevalence estimation — Nature Communications (2025)
Selected Research & Projects
Selected work in statistical identifiability, Bayesian inference, simulation, and uncertainty quantification.
Identifiability in Latent-Variable Models
Study of identifiability in latent-variable models under heterogeneous observation designs using Jacobian rank, null-space analysis, and experimental-design structure to characterize when prevalence and diagnostic-accuracy parameters can be uniquely recovered.
Bayesian latent class modeling with a continuous latent severity variable for estimating disease prevalence and diagnostic test accuracy without a gold standard, with emphasis on model reduction, parameter recovery, and uncertainty quantification.
Hierarchical Global Prevalence Estimation
Bayesian hierarchical estimation of disease prevalence across studies, regions, and countries by combining heterogeneous diagnostic study designs with imperfect-test information and uncertainty in diagnostic accuracy.
Uncertainty Propagation for Global Economic Loss
Propagating posterior uncertainty in disease prevalence together with uncertainty in livestock populations and economic inputs to estimate country-level and global distributions of economic loss.
Analysis of randomized product experiments including metric definition, power and sample-size calculations, treatment-effect estimation, uncertainty quantification, and experiment diagnostics.
Experience
- Study identifiability in latent-variable models under heterogeneous observation designs using analytical and computational methods.
- Develop Bayesian hierarchical models for prevalence estimation and diagnostic testing without a gold standard.
- Simulations to evaluate parameter recovery, robustness, and uncertainty quantification.
- Develop latent-variable extensions that reduce model complexity and improve estimation in sparse-data settings.
- Teach foundational probability concepts including random variables, discrete and continuous distributions, expectation, variance, conditional probability, independence, and normality.
- Teach statistical reasoning through sampling distributions, estimation, hypothesis testing, regression, and computational analysis in R.
- Developed and deployed credible intervals for predictive models to quantify uncertainty.
- Designed repeatability and reproducibility studies to quantify measurement variability and support product validation.
- Built predictive models relating chemical composition to transmission-fluid performance to support product-development decisions.
- Applied high-dimensional regression and clustering methods to DNA methylation data, identifying biomarkers linked to disease pathways.
- Analyzed NHANES data to evaluate dietary risk factors for hypertension in pregnancy, contributing to peer-reviewed publications.
Publications
Nature Communications, 16:11600 (2025)
BMC Biology, 21:30 (2023)
Nutrients, 14(12):2545 (2022)