Abstract: Causal mediation analysis plays a central role in uncovering mechanistic pathways that clarify how causal relationships operate through intermediate variables. This type of analysis is widely applied across scientific disciplines, yet numerous challenges remain. These include translating a hypothesized directed acyclic graph (DAG) topology into a generative algebraic model, assessing identifiability conditions via sensitivity analysis, evaluating spillover effects in socialnetwork or spatial correlated setting, and modeling complex mediators. A pressing need exists for new statistical frameworks and software tools capable of handling diverse data types, such as high-dimensional, functional, metric-space, and imaging data. In this talk, I will review several open and urgent problems in causal mediation analysis that call for innovative solutions from the statistical communit.

Bio: Dr. Song is Professor of Biostatistics at the University of Michigan School of Public Health, Ann Arbor. He received his PhD in Statistics from the University of British Columbia, Vancouver, Canada in 1996. He has published over 250 peer-reviewed papers and graduated 28 PhD students and trained 6 postdoc research fellows. Dr. Song's current research interests include data integration, distributed inference, high-dimensional data analysis, longitudinal data analysis, mediation analysis, and spatiotemporal modeling. He is IMS Fellow, ASA Fellow and Elected Member of the International Statistical Institute. Dr. Song now serves as Area Editor of the Annals of Applied Statistics (Medicine, EHR and Smart Health), Associate Editor of the Journal of American Statistical Association, Journal of the Royal Statistical Society Series B (Statistical Methodology) and the Journal of Multivariate Analysis.