Guest lecture
In this talk, Dr. Terry Therneau will discuss key challenges in modeling cognitive aging using both direct cognitive measures (such as memory tests and activities of daily living) and biomarkers that assess underlying pathological processes (including amyloid PET and serum assays).
A central difficulty is the mismatch between long disease timelines—often spanning 20 years or more—and relatively short observation windows of 4–6 years in most longitudinal studies. Additional challenges arise from substantial between-subject variability in age of onset (e.g., APOE ε4 status shifting amyloid onset by approximately six years), irregular measurement availability across studies, low signal-to-noise ratios for many markers, and informative censoring.
The talk will explore the goal of integrating information across multiple markers, ages, and patients to better understand how underlying pathologies develop and interact. Dr. Therneau will highlight modeling strategies such as multi‑marker nonlinear mixed effects models and hidden Markov models, discussing both their promise and computational challenges.
We warmly encourage faculty, students, and researchers with interests in public health sciences, aging research, statistics and epidemiology, to attend. All are welcome!
Biography
Dr. Terry Therneau
Dr. Therneau earned his PhD in Statistics from Stanford University in 1983 and has spent his distinguished career at the Mayo Clinic. His research has had a major impact across numerous health domains, including cancer, cardiovascular disease, transplantation and neurological disorders.
He is internationally recognized for his contributions to survival analysis, including foundational work on frailty models, robust inference, influence analysis, and goodness‑of‑fit methods. Dr. Therneau co‑authored the influential book Modeling Survival Data: Extending the Cox Model (Springer, 2000) with Dr. Patricia Grambsch, and he maintains the widely used survival package in R, which is central to survival analysis in public health research.