Biostatistics Seminar: Laurel Beckett
DATE: Tuesday, February 18th, 2014
TIME: 4:10pm (refreshments at 3:30pm, MSB 4110)
LOCATION: Mathematical Sciences Building 1147
SPEAKER: Laurel Beckett, Division of Biostatistics, School of Medicine, University of California Davis
TITLE: Mixture models for sequences based on censored data
ABSTRACT:
The Alzheimer’s Disease Neuroimaging Initiative (ADNI) and similar studies now provide an increasing resource of longitudinal data about cognition, function, neuroimaging, and fluid biomarkers, across the spectrum from normal controls through mild impairment to dementia. Jack (2010) has proposed a sequence for the onset of abnormalities in different domains in Alzheimer’s disease (AD). This sequence, however, may not be uniform in all people. Other causes of dementia may have different patterns of progression, and even AD may not be uniform, due either to co-existing pathologies or to heterogeneity of the disease. We propose a model for mixtures of sequences and an approach to estimation from partially ranked data like that from longitudinal cohorts like ADNI, in which some transitions have already occurred or are not observed yet in some participants. We show that mixtures can be detected in such data, allowing us to describe and quantify heterogeneous sequences. If time permits, we will also show applications to other medical and non-medical datasets.
AUTHORS: This is joint work of Laurel Beckett (PhD) and Erik Gregory.