This project encompasses a wide scope of statistical collaboration with neuroscientists and clinical neurologists,and the developoment of statistical methods to address problems encountered in collaborative studies. The focus of collaborative research is on statistical planning and design of experiments, statistical analysis and modeling of data, and statistical inference. Examples of current studies include: prevalence studies of neurological diseases in the U.S., China and Spain; case-control study of ultra microassay of neonatal blood constituents for increased risk of autism; analysis of the relationship of cytokines, neurotrophins and growth factors to cerebral palsy and autism risk; clinical trial of the effectiveness of magnesium sulfate for the prevention of white matter lesions in infants from high-risk pregnancies; statistical modeling of changes in PET imaging of patients with epilepsy; identification of clinical factors known at hospital admission that are associated with outcome of ischemic stroke; design and analysis of Phase I/II clinical trials for the evaluation of stroke treatment; population prevalence studies of epilepsy; analysis of developmental brain differences between childhood onset schizophrenic patients and healthy controls; longitudinal modeling of gray/white matter development in healthy children; longitudinal MRI study of corpus callosum development in healthy children; statistical models for spectroscopic imaging of the brain; analysis and statistical modeling of mechanisms of use-dependent plasticity in human motor cortex; the relationship of seizure frequency to hippocampus volume in temporal epilepsy; and post operative changes in cerebral metabolism in temporal lobe epilepsy. Statistical research usually derives from problems encountered in collaborative studies. Areas of research include: population screening methods for disease; establishment of reference range values for diagnostic tests; time series models for longitudinal imaging data; Bayesian designs for Phase I/II clinical trials with control of both efficacy and adverse events; quantification of selection bias in surveys; nonlinear response models for nondecreasing, time-dependent multiple compartment systems; and new likelihood tests for logistic mixture models in time series.
Showing the most recent 10 out of 21 publications
