The panel members are:
Then the results of a preliminary investigation into the causes of overdispersion in log-linear models will be presented, together with a literature review. Possible directions for further research will be discussed.
This is joint work with Jennifer Ashmore & Doug Ramsay (Dept of Orthodontics), Brian Leroux, & Paul Sampson.
The aim of this study is to describe the movement of molars within the jaw for three groups of 7-12 year old children with class II malocclusion: a control group with no orthodontic treatment, and two applications of headgear treatment. All subjects have had dental casts taken of the maxilla (upper jaw) at the beginning of the study and 2 years later. One of the treatment groups had a series of casts taken at 2 month intervals.
The data are 3-D points digitized from the casts using a "Microscribe".
The data analysis consists of three stages:
1) Orientation Rotation and translation of time 1 data so that movement
in the X, Y, Z directions (and angles of rotation) have the desired interpretations.
2) Superimposition Data for subsequent casts is superimposed on the
first cast's data by a Procrustes least-squares fit of homologous points.
3) Analysis Determine how the molars have moved over time. We are interested
in both translation (movement along the X, Y, or Z axis) and rotation.
Stages 1) and 2) are more or less complete, and will be described for the group. We are still working to develop methods for the analysis, so that will be the focus of the talk.
This research was motivated by a specific dataset from the Cardiovascular Health Study (CHS). Specifically, 3647 CHS participants had an MRI scan, from which the location of MRI-detected strokes were recorded in terms of a 23 region atlas of the brain. The goal of the study is to assess the magnitude of association between strokes in these locations and various response variables such as measures of cognitive function, depression, or incident events. A difficulty with these data is that there are a large number of regions, and many with sparse data. To simplify presentation and improve estimation we would like to combine the regions to form a reduced atlas of the brain.
To address this problem I have developed a clustering algorithm tailored to the features of the MRI data. Regions are combined based on their association with the response, rather than with each other. The algorithm can adjust for potentially confounding variables during the clustering process. The statistical properties and performance of the algorithm were evaluated via simulation studies. Specifically, we considered whether the algorithm successfully captured true underlying structure, whether good parameter estimates were obtained, and whether there is a benefit to clustering over simply using all the regions individually. Various noise levels and sample sizes were studied. Applications which consider a range of possible scenarios will also be presented.
I will present graphical and numerical methods for checking the adequacy of the logistic regression model for matched case-control data. The proposed methods are based on the cumulative sum of residuals over the covariate or linear predictor. Under the assumed model, these cumulative residual processes converge weakly to zero-mean Gaussian processes, whose distributions can be approximated via simulations. The observed cumulative residual pattern can then be compared both visually and analytically to a number of simulated realizations from the approximate null distribution. These methods are useful in checking the functional form of each covariate, the relative risk function as well as the overall model adequacy. The performance of the proposed methods was assessed through simulation studies. Illustration with real datasets will be provided.
The near completion of the Human Genome Project and breakthroughs in
the technology for discovering and typing DNA sequence variants has offered
the potential to open wide new windows into the study of genome complexity.
In this talk, I will first overview the current paradigm of Genetic Epidemiology
and describe challenges that may face the field. If time permits, I will
present one of my research topics, candidate gene association analysis
using correlated failures times as outcomes.
Last Modification: 08 December 2000