Research

We develop statistical inference theory for complex data structures that arise in modern biomedical and health studies, where standard regularity conditions often fail. Our work focuses on likelihood and pseudolikelihood based inference under nonstandard settings, including boundary constraints, latent variable models, nonconvex parameter spaces, and irregular or informative observation processes. We study identifiability, asymptotic behavior of test statistics, and valid hypothesis testing in settings where classical likelihood theory breaks down. These theoretical advances provide a principled foundation for reliable inference in diagnostic accuracy studies, longitudinal data with informative visit times, and semiparametric models with boundary problems.

Selected papers:

• Wang, J. Y., Ye, Z. S., & Chen, Y. (2024). Likelihood-based inference under nonconvex boundary constraints. Biometrika, 111(2), 591-607.
• Duan, R., Cao, M., Ning, Y., Zhu, M., Zhang, B., McDermott, A., … & Chen, Y. (2020). Global identifiability of latent class models with applications to diagnostic test accuracy studies: A Gröbner basis approach. Biometrics, 76(1), 98-108.
• Chen, Y., Ning, J., Ning, Y., Liang, K. Y., & Bandeen-Roche, K. (2017). On pseudolikelihood inference for semiparametric models with boundary problems. Biometrika, 104(1), 165-179.
• Hong, C., Ning, Y., Wang, S., Wu, H., Carroll, R. J., & Chen, Y. (2017). PLEMT: A novel pseudolikelihood-based EM test for homogeneity in generalized exponential tilt mixture models. Journal of the American Statistical Association, 112(520), 1393-1404.
• Chen, Y., Ning, J., & Cai, C. (2015). Regression analysis of longitudinal data with irregular and informative observation times. Biostatistics, 16(4), 727-739.
• Chen, Y., & Liang, K. Y. (2010). On the asymptotic behaviour of the pseudolikelihood ratio test statistic with boundary problems. Biometrika, 97(3), 603-620.