Classification Trees and Covariates: Applications in Neuroscience

• Activity Number: 4
• Type: Invited
• Date/Time: Sunday, August 3, 2014 : 2:00 PM to 3:50 PM
• Sponsor: Memorial
• Author(s): Josephine Asafu-Adjei and Allan Sampson*+
• Companies: Harvard School of Public Health and University of Pittsburgh
• Keywords: classification tree ; covariate ; biomarker

Abstract

Studies in post-mortem brain tissue are an integral resource in identifying neurobiological markers that differ for subjects with a mental disorder compared to normal controls. In addition to the measured multiple biomarkers, certain covariates are collected such as post-mortem interval (PMI), i.e., elapsed time from death to tissue collection, and tissue storage time (TST), i.e., the time length tissue is stored. Classification trees are a valuable tool to identify the biomarkers and their relationships that distinguish the disorder from normal controls. However, a subject's PMI and TST while not relevant to disorder characterization can impact a subject's biomarker measurements. In this talk we present several approaches to classification trees that suitably account for any impacts of covariates, e.g., PMI or TST. Our approaches are first developed in general parametric and semi-parametric population settings and based on these, appropriate parameter estimators are obtained. An illustrative application to neuroscience will be given.