Prof. Mohammad Jafari Jozani (University of Manitoba) - (SPECIAL SEMINAR)

Title: Developing a New Rank-Based Quantile Regression Methodology for Monitoring the Prevalence of Osteoporosis.

In osteoporosis studies, quantiles of the distribution of bone mineral density (BMD) and their relationship with risk factors such as age, family history, gender, etc., are important. Low BMD is associated with higher probability of leg and pelvis fractures due to falls and causes significant public health issues, especially in elderly women, leading to high medical cost, inability to live independently, and even risk of death. In this study, we use a large cohort study in the Canadian province of Manitoba and develop a new methodology to monitor the prevalence of osteoporosis using more efficient and cost-efficient follow-ups. We propose to use expert knowledge and/or suitable measurements from the base-line study to design follow-up sample selection procedures that are based on ranks. We obtain the ranks by comparing a patient with a small number of other patients, randomly selected from the study cohort. We propose a new check-function to incorporate the rank information associated with selected (nominated) samples in the estimation process. Strategies are given to design proper follow-up studies for a given population quantile. Numerical studies show that rank-based quantile regression models are more efficient than those based on simple random sampling (SRS) for analyzing upper/lower tail quantiles of the distribution of BMD. Among other results, we observe that in some cases, our proposed method requires about one‐tenth of the sample used in SRS to estimate the lower/upper tail conditional quantiles with comparable mean squared errors. This is a dramatic reduction in time and cost compared with the usual approach and is very efficient to design better follow-ups in large cohort studies.