|
Activity Number:
|
253
|
|
Type:
|
Topic Contributed
|
|
Date/Time:
|
Tuesday, August 5, 2003 : 12:00 PM to 1:50 PM
|
|
Sponsor:
|
Section on Statistics in Epidemiology
|
| Abstract - #302170 |
|
Title:
|
Modeling Defects in the Reward System in the Human Brain
|
|
Author(s):
|
Raymond G. Hoffmann*+ and Paul Hoffmann and Robert Risinger and Elliot Stein
|
|
Companies:
|
Medical College of Wisconsin and Marquette University and Medical College of Wisconsin and National Institutes of Health
|
|
Address:
|
Div. of Biostatistics, Milwaukee, WI, 53226-0509,
|
|
Keywords:
|
neural network ; Bayesian neural network ; fMRI ; biomathematical models
|
|
Abstract:
|
Connectionist neural network models have been successful in testing the possible causes of Schizophrenia (Cohen and Servan-Schrieber) and deep dyslexia (Sejnowski). A connectionist model of the brain is usually implemented with a set of interconnected neural networks (Nnet). Each Nnet substructure is used to model the function of a particular structure of the human brain. Lesions (broken connections), signal masking noise (poor connections or poor function of a structure) and redundant pathways (additional pathways that impair normal functioning) are used to test hypotheses about the cause of malfunctions in the normal operation of the brain. The model in this study is based on the structures of the reward system so that (1) hypothesized changes that resulted from nicotine addiction and (2) hypothesized differences that can enhance nicotine cessation can be tested on a model system. FMRI data from the effect of nicotine abstinence, both short-term and long-term, are used to estimate the parameters of the model. Alterations in the model that produce changes in the predicted signal are compared with changes observed in fMRI times series from real and simulated subjects.
|
