T lymphocytes may be classified as naive or memory cells, depending on
whether they possess immunological memory. The central tenet of this
work is that memory lymphocytes revert to naive lymphocytes in vivo. T
his phenomenon is modeled by a mechanism qualitatively similar to Demo
ivre's law of human mortality. Additionally, both lymphocyte subsets u
ndergo decay. Model parameters and their standard errors are estimated
by maximizing a likelihood function constructed by Kalman filtering o
f the experimental data on the assumption that it is contaminated by m
easurement noise of constant relative error. (C) Elsevier Science Inc.
, 1997.