Department Seminar by Teng Fei

Please Note: This seminar will be given online.

Latent Class Methods for Complex Chronic Disease Data

Latent class analysis is an intuitive and powerful data-driven tool to characterize the heterogeneity of chronic disease phenotypes. Motivated by the different research questions from Alzheimer’s disease research, we have developed a series of latent class methods which can overcome the various limitations of existing methods, such as estimation bias, restrictive parametric model assumptions, and expensive computation. In this talk, I will focus on presenting our proposed structural time-dependent competing risks model, which is novelly and sensibly formulated to assess the association between latent classes of baseline cognitive performance in patients with mild cognitive impairment (MCI) and their subsequent neuropathological features. I will discuss the associated estimation and inference procedures as well as numerical studies including an application to an MCI cohort from the National Alzheimer’s Coordinating Center (NACC) data. In addition, I will provide a brief overview of our developments of robust semi-parametric latent class methods for longitudinal data and survival data.