Jianbin Tan | Duke University

Welcome to Jianbin's homepage!

I am currently a Postdoctoral Associate in the Department of Biostatistics & Bioinformatics, Duke University, supervised by Prof. Anru Zhang and Prof. Pixu Shi. I received my PhD in Statistics from Sun Yat-sen University in 2023, advised by Prof. Hui Huang. I won the Outstanding Doctoral Thesis Award at Sun Yat-sen University and the 2026 IMS New Researcher Travel Award.

My research interests lie in developing statistical and AI methods for modern complex data, supporting effective analysis in health, biological, and epidemiological sciences.

AI and Statistics for Health

We develop AI and statistical methods for health-related data with complex sampling schemes, with an emphasis on generative modeling, representation learning, structure-aware imputation, and reliable prediction and inference.

Representative publications:

Smooth Flow Matching for Synthesizing Functional Data Preprint

Functional-SVD for Heterogeneous Trajectories: Case Studies in Health JASA

Integrated Analysis for Electronic Health Records with Structured and Sporadic Missingness JBI

Associating High-Dimensional Longitudinal Datasets through an Efficient Cross-Covariance Decomposition Preprint

Epidemiologic Data Analysis

We develops statistical methods for dynamic epidemiologic and public health data, including causal inference, transmission modeling, and association discovery, to characterize policy effects and enable reliable inference and public health evaluation.

Representative publications:

The Effects of Stringent and Mild Interventions for Coronavirus Pandemic JASA

Transmission Roles of Symptomatic and Asymptomatic COVID-19 Cases: A Modelling Study I&E

Functional Clustering for Longitudinal Associations between Social Determinants of Health and Stroke Mortality in the U.S. AOAS

Age-Related Model for Estimating the Symptomatic and Asymptomatic Transmissibility of COVID-19 Patients BIOM

Functional/Longitudinal Data

This research develops statistical learning methods for functional/longitudinal data under complicated settings. We emphasize model-free approaches that provide structural adaptivity and accommodate complex sampling schemes, which are key challenges in modern functional and longitudinal data analysis.

Representative publications:

Smooth Flow Matching for Synthesizing Functional Data Preprint

Functional-SVD for Heterogeneous Trajectories: Case Studies in Health JASA

Graphical Principal Component Analysis of Multivariate Functional Time Series JASA

Functional Clustering for Longitudinal Associations between Social Determinants of Health and Stroke Mortality in the U.S. AOAS

Associating High-Dimensional Longitudinal Datasets through an Efficient Cross-Covariance Decomposition Preprint

Dependence-Adaptive Modelling for Functional Time Series Preprint

Statistical Learning of Differential Equations

This research focuses on developing statistical methods for differential equation modeling and learning, including parameter estimation and vector field inference, as well as differential-equation–based generative modeling and other related tasks.

Representative publications:

Green’s Matching: an Efficient Approach to Parameter Estimation in Complex Dynamic Systems JRSSB

Smooth Flow Matching for Synthesizing Functional Data Preprint

Age-Related Model for Estimating the Symptomatic and Asymptomatic Transmissibility of COVID-19 Patients BIOM

Sparse Equation Matching: a Derivative-Free Learning for General-Order Dynamical Systems Preprint