Biography

I am a Postdoctoral Researcher in the Department of Biostatistics, Epidemiology and Informatics at the University of Pennsylvania, working with Prof. Hongzhe Li. I received my Ph.D. in Statistics from Fudan University, advised by Prof. Zhongyi Zhu, and was a visiting Ph.D. student at the University of California, Irvine, hosted by Prof. Annie Qu. My research develops statistical methods for learning and inference from high-dimensional and heterogeneous data, with an emphasis on their reliable application in modern biology and emerging AI-enabled scientific research.

Research

My research interests span AI-augmented inference, data integration and transfer learning, and causal inference, with particular emphasis on their intersections with integrative genomics and single-cell genomics. A central theme of my current work is statistical learning and inference under heterogeneity, pursued through three complementary directions: borrowing shared information across heterogeneous data, exploiting heterogeneity for structural identification, and learning individual differences through shared structure.

Information Borrowing across Heterogeneous Data Sources Information Borrowing across Heterogeneous Data Sources How can information from AI-generated and other heterogeneous data sources be used reliably without imposing inappropriate sharing? I develop methods that learn when and how much information to borrow from related but possibly imperfect data sources, while protecting against inappropriate sharing. These methods are applied to AI-generated biological data, particularly virtual cells, and to integrative genomic studies across biological contexts. View related work Hide related work

Related work

Calibrating auxiliary information from AI-generated or observational data

Generation-Powered Inference for Distribution-valued Outcomes Yijiao Zhang and Hongzhe Li. arXiv preprint, 2026. [arXiv]
A Data Fusion Method for Quantile Treatment Effects Yijiao Zhang and Zhongyi Zhu. Statistica Sinica, 35, 981–1002, 2025. [Paper] [arXiv]

Adaptive information transfer across related sources

Transfer Learning for High-dimensional Quantile Regression via Convolution Smoothing Yijiao Zhang and Zhongyi Zhu. Statistica Sinica, 35, 939–958, 2025. [Paper] [arXiv]
Transfer Learning for High-dimensional Quantile Regression with Distribution Shift Ruiqi Bai, Yijiao Zhang, Hanbo Yang, and Zhongyi Zhu. arXiv preprint, 2026. [arXiv]
Covariate-Elaborated Robust Partial Information Transfer with Conditional Spike-and-Slab Prior Ruqian Zhang, Yijiao Zhang, Juan Shen, Zhongyi Zhu, and Annie Qu. Journal of the American Statistical Association, to appear, 2026. [Paper]
Optimal Transport based Cross-Domain Integration for Heterogeneous Data Yubai Yuan, Yijiao Zhang, Babak Shahbaba, Norbert Fortin, Keiland Cooper, Qing Nie, and Annie Qu. Journal of the American Statistical Association, 120(551), 1449–1462, 2025. [Paper]
Exploiting Heterogeneity for Causal and Invariant Learning Exploiting Heterogeneity for Causal and Invariant Learning How can variation across interventions and environments become a source of structural identification? I develop methods that use variation across interventions and environments to identify causal structures and stable mechanisms that are difficult to recover from a single environment. These methods are applied to gene regulatory network learning and single-cell genomics. View related work Hide related work

Related work

Causal Network Discovery from Interventional Count Data with Latent Linear DAGs Yijiao Zhang and Hongzhe Li. arXiv preprint, 2026. [arXiv]
Bayesian Environment Invariant Regression Ruqian Zhang, Juan Shen, and Yijiao Zhang. arXiv preprint, 2026. [arXiv]
Individualized Learning with Shared Structure Individualized Learning with Shared Structure How can shared structure improve learning while preserving meaningful individual differences? I develop methods that borrow shared latent structure across individuals while preserving meaningful individual-specific variation. These methods are applied to personalized epigenomic and biomedical analysis, as well as personalized language generation. View related work Hide related work

Related work

Individualized Dynamic Mediation Analysis Using Latent Factor Models Yijiao Zhang, Yubai Yuan, Yuexia Zhang, Zhongyi Zhu, and Annie Qu. Journal of the American Statistical Association, to appear, 2026. [Paper]
Syntax-Guided Diffusion Language Models with User-Integrated Personalization Ruqian Zhang, Yijiao Zhang, Juan Shen, Zhongyi Zhu, and Annie Qu. The Annals of Applied Statistics, to appear, 2026. [arXiv]