A115 Crabtree Hall
Meredith L. Wallace of Pitt's Department of Psychiatry presents a talk entitled "Variable Selection for Skewed Model-Based Clustering: Revealing Novel Sleep Phenotype" as part of the Biostatistics 2016 Spring Seminar Speaker Series.
Visit publichealth.pitt.edu/calendar for details.
Variable Selection for Skewed Model-Based Clustering: Revealing Novel Sleep Phenotype
ABSTRACT: In sleep research, applying mixture models to sleep characteristics captured through multiple data types, including self-reported sleep diary, a wrist monitor capturing nighttime movement (actigraphy), and brain waves (polysomnography), may suggest new phenotypes that reflect actual disease mechanisms rather than self-reported symptoms. However, a direct mixture model application is challenging because: 1) there are many sleep characteristics from which to choose, 2) sleep measures are often highly skewed, even in homogenous samples, and 3) solutions that are interesting clinically often incorporate all three data types. Dr. Wallace will discuss her novel variable selection method that is based on the restricted multivariate skew-normal distribution and which suggests multiple statistically plausible solutions incorporating all data types. She will demonstrate her proposed methods using a sample of older adults and reveal novel sleep phenotypes to help researchers generate hypotheses regarding pathophysiological mechanisms and new personalized treatments to be developed.
Biostatistics 2016 Spring Seminar Speaker Series presents
Meredith L. Wallace, PhD
Assistant Professor, Department of Psychiatry, University of Pittsburgh
Thursday, January 21, 2016
3:30 pm, A115 Crabtree Hall