Seminar Series
Title
Update on the Canadian Longitudinal Study on Aging (CLSA)
Speaker
Dr. Lindsay Wallace
Time and Date
12 – 1 pm, Thursday, Oct 9, 2025
Location
MS Teams Join the meeting now
Bio
Dr. Lindsay Wallace is an Assistant Professor in Community Health and Epidemiology at Dalhousie University and a co-Principal Investigator of the Dalhousie Canadian Longitudinal Study on Aging site. She completed an MSc at McGill University in Neuroscience, a PhD from Dalhousie in Interdisciplinary Health Studies, and a Post-Doctoral Fellowship at the University of Cambridge in Public Health. The focus of Dr. Wallace’s research is understanding the physical, social, and structural conditions that give rise to chronic disease, and how we can implement scalable solutions to reduce morbidity. The majority of her research to date has examined the interplay between frailty and dementia in this context. Dr. Wallace’s work is inherently interdisciplinary and blends insights from epidemiology, public health, and neuroscience to address the complexity of aging and disease. She analyses data from large longitudinal cohort studies to support her research, as well as using interventional techniques and policy analysis.
Synopsis
The Canadian Longitudinal Study on Aging (CLSA) is a large, national research platform that follows more than 50,000 Canadians between the ages of 45 and 85 at baseline to better understand the aging process and its impacts on health, social, and economic factors over time. Participants are followed for up to 20 years. Data are collected every 3 years through questionnaires (on health, lifestyle, social, and economic factors), in-depth physical assessments at data collection sites, and biological samples (like blood and urine) for biomarker and genetic analyses. Data collection happens across 11 sites in 7 provinces, making it one of the largest and most detailed aging studies in the world. In this presentation, Dr. Wallace will review the platform, including information on data availability, access, and research impact, as well as focus on its utility for student researchers, as well as the resources available to support its use.
----------------------------------------------------------------------------------------------
Title
A Copula Joint Model to Address Endogeneity in the Relationship between ED Length of Stay and Readmission
Speaker
Dr. Guanjie Lyu
Time and Date
12 – 1 pm, Thursday, Nov 6, 2025
Location
MS teams: Join the meeting now
Bio
Dr. Guanjie Lyu is a Postdoctoral Fellow in the Department of Community Health and Epidemiology at Dalhousie University, working under the supervision of Dr. Cindy Feng. He earned his PhD in Statistics from the University of Windsor in 2024. Dr. Lyu’s research focuses on Bernstein polynomials and copula modeling, with broad applications in statistics and data analysis. His work spans both methodological development and applied problems, with a particular interest in dependence modeling and joint modeling of complex health data.
Professional Profile: https://guanjielyu.github.io/
Synopsis
The complex relationship between a patient's emergency department (ED) length of stay (LOS) and their subsequent ED readmission risk is a significant challenge in health services research. This presentation will show how to address the endogeneity of LOS, where unobserved factors can bias traditional models. We propose a novel framework using a copula joint model to simultaneously investigate this relationship and account for endogeneity. Our approach uses a continuous smooth function to capture the non-linear effect of LOS, revealing a sharply increasing risk for very short stays that then plateaus for intermediate stays. We validate our model using real-world ED data from Nova Scotia, Canada. A comprehensive simulation study compares our copula joint model against separate and two-stage regression approaches. The results demonstrate the superior performance and more accurate parameter estimates of the copula model, highlighting its power as a flexible statistical tool for analyzing outcomes with complex dependencies and endogeneity.
----------------------------------------------------------------------------------------------