Provided by Cognitive Sciences ePrint Archive
J. Medical Informatics and Technologies,
5 (2000) 27-34
*
Wlodzislaw DUCH
THERAPEUTIC IMPLICATIONS OF COMPUTER MODELS OF BRAIN
ACTIVITY FOR ALZHEIMER DISEASE.
Neural models of large-scale brain processes help to explain many features of neuropsychological syn-
dromes and psychiatric disease. Two associative memory models useful to understand some aspects of cognitive
impairments in Alzheimer disease are discussed. The first model is based on the synaptic deletion and compensa-
tion while the second on the synaptic runaway phenomenon. The models seem to be complementary, explaining
different types of Alzheimer disease. They allow to draw several therapeutic suggestions that may help to slow
down the development of the disease in its early stages.
1. Introduction
Neural networks and other computational intelligence models inspired by our under-
standing of the brain are widely used for medical diagnostics support, signal and image
analysis, monitoring, search for carcinogenic agents and other data analysis tasks. In these
applications neural networks compete with statistical, machine learning and other mathe-
matical techniques. A qualitatively different area of neural modeling focuses on understand-
ing of physiological responses of single neurons or small groups of neurons. Sophisticated
biophysical models of compartmental neurons provide information directly related to neu-
rophysiological parameters measured in experiments. Already in 1994 Callaway and col-
laborators modeling reaction times to different drugs stated: “Neural network models offer a
better chance of rescuing the study of human psychological responses to drugs than any-
thing else currently available” [3]. Classical methods of psychiatry and neuropsychophar-
macology are restricted to observations of correlations between behavior and physiological
responses of the organism to medical treatments. They do not provide any insights into the
mechanisms leading to neuropathological behavior at the neural level. Simulations may
provide understanding of neural responses to biochemical substances acting at the ionic
channel levels.
A third area of neural modeling that slowly grows in importance concerns understand-
ing of the large-scale processes going on in the brain. Brain processes are very complex and
therefore neural networks based on biophysical, spiking neural models cannot be used for
large-scale modeling. Simulations should capture casual relations between activity of brain
Katedra Metod Komputerowych, Uniwersytet Mikolaja Kopernika, ul. Grudzi^dzka 5, 87-100 Torun, Poland
http://www.phys.uni.torun.pl/~duch