Holder： Zhonghua Liu(Columbia University)
Time：2026-01-05 10:00-11:15
Location：Conference Room 220, New Public Health Building, Peking University Health Science Center
Abstract:
Causal inference in large-scale biobanks is complicated by right-censored outcomes, weak or invalid genetic instruments, and complex high-dimensional omics data. In this talk, I will introduce new robust Mendelian randomization (MR) methods that enable valid causal inference for censored time-to-event outcomes in the UK Biobank. The framework integrates semiparametric identification, generalized empirical likelihood, and modern deep learning to handle many weak and pleiotropic instruments while accounting for non-Neyman orthogonal nuisance components. I will demonstrate how these advances recover attenuated effects in standard AFT models and yield biologically interpretable causal estimates for cardiometabolic and neurological diseases. In the second part, I will show how proteomics and AlphaFold3-based structural predictions can be combined with MR to identify causal proteins and accelerate drug discovery for Alzheimer's disease and other conditions. Together, these methods illustrate how statistical innovation and AI-driven biology can jointly advance precision public health.
About the Speaker:
Dr. Zhonghua Liu is an Assistant Professor of Biostatistics at Columbia University and a member of the Data Science Institute. Before joining Columbia, he was an Assistant Professor in the Department of Statistics and Actuarial Science at the University of Hong Kong and a Quantitative Strategist at Morgan Stanley in New York. He received his doctorate in Biostatistics from Harvard University, where he was a Harvard Presidential Scholar, under the supervision of Professor Xihong Lin. Dr. Liu’s research lies at the intersection of causal inference, artificial intelligence, and genomics. He develops semiparametric, robust, and deep learning methods for high-dimensional data, integrating statistical theory with modern deep learning to advance causal discovery and inference in precision medicine. His recent work has appeared in JASA, JRSS-B, Biometrika, Biometrics, AOAS, NeurIPS, ICML, Nature Computational Science, and Cell Genomics.
Your participation is warmly welcomed!
欢迎扫码关注北大统计科学中心公众号，了解更多讲座信息!
