Statistical methods for study of suicidal behavior

New York State Psychiatric Institute, New York NY

Investigators

Linked publications & trials

Abstract

This project aims at developing cutting edge statistical and computational methods applicable to a wide variety of neuroscience problems, in particular, to suicide research. It consists of two related sub-projects. The first one develops methods for analyzing brain images, such as autoradiographic images of postmortem tissue and other neurochemical maps (e.g., "region of interest" data for PET images). The other subproject is an exploratory investigation of advanced techniques for modeling brain function and disease. The subprojects are linked because models of the brain can be used in analyzing neurochemical map data. An examination of brain regional neurochemistry reveals differences in measures such as receptor binding among brain regions. The anatomical distribution of neurotransmitter receptors is generally studied using quantitative autoradiography. Autoradiographic studies in complete coronal sections of human brain like those carried out by the Human Neurobiology Core are extensive to perform and human tissue is previous. PET images of live subjects such as those produced by the Brain Imaging Core are expensive to perform and human tissue is previous. PET images of live subjects such as those produced by the Brain Imaging Core are also expensive. To make best use of these valuable data, it is important to use powerful statistical methods. Thus, it makes sense to examine binding in a number of different brain regions. However, it is difficult to obtain and assay large numbers of postmortem samples or to image a large number of live subjects. This creates challenging multiple comparisons problems. A potentially powerful way to handle these multiple comparison problems involves use of mathematical models of the dependence among the brain regions of interest. In addition to the large array of standard statistical models one might use, in recent years, "computational neuroscientists" have put intensive effort into developing sophisticated mathematical and/or computer models of brain function and disease. The application of such modeling techniques in suicide research will be explored. The databases in the Conte Center for the Neuroscience of Mental Disorder will be used to fit and test these models.