2022/12/27
Bayesian mixed models for longitudinal data
演講相關內容如下：
===========================================
時間：12/30（五）13:10-15:30
地點：社科院北棟2樓心理系階梯教室
講題：Bayesian mixed models for longitudinal data
講者：李國榮老師（成功大學統計系副教授）
個人網站：https://sites.google.com/view/kuojunglee
摘要：Longitudinal studies have been conducted in a wide variety of application areas, including psychology, medicine, economics and social sciences. In such studies, longitudinal data are collected over time, and thus repeated outcomes within each subject may therefore tend to be serially correlated. To account for both the serial correlation within subjects and the specific variability between subjects, we propose several Bayesian random-effects model to analyze longitudinal data using the hypersphere decomposition approach to solve the positive-definiteness constraint and high-dimensionality of the correlation matrix. In this talk, I present the models through real examples:
1. A nonalcoholic fatty disease dataset to study correlations of responses across time to explain the joint variability of lung functions and body mass index in multiple responses over time.
2. A cohort study for metabolic syndrome from Korean Genomic Epidemiology Study (KGES).
3. A lung cancer dataset to compare the cancer response rates (RR) in the two arms.
1. A nonalcoholic fatty disease dataset to study correlations of responses across time to explain the joint variability of lung functions and body mass index in multiple responses over time.
2. A cohort study for metabolic syndrome from Korean Genomic Epidemiology Study (KGES).
3. A lung cancer dataset to compare the cancer response rates (RR) in the two arms.
===========================================