TY - JOUR
T1 - Exploiting the bootstrap method for quantifying parameter confidence intervals in dynamical systems
AU - Joshi, M.
AU - Seidel-Morgenstern, A.
AU - Kremling, A.
N1 - Funding Information:
The financial support by the DFG (FOR 447) is gratefully acknowledged.
PY - 2006/9
Y1 - 2006/9
N2 - A quantitative description of dynamical systems requires the estimation of uncertain kinetic parameters and an analysis of their precision. A method frequently used to describe the confidence intervals of estimated parameters is based on the Fisher-Information-Matrix. The application of this traditional method has two important shortcomings: (i) it gives only lower bounds for the variance of a parameter if the solution of the underlying model equations is non-linear in parameters. (ii) The resulting confidence interval is symmetric with respect to the estimated parameter. Here, we show that by applying the bootstrap method a better approximation of (possibly) asymmetric confidence intervals for parameters could be obtained. In contrast to previous applications devoted to non-parametric problems, a dynamical model describing a bio-chemical network is used to evaluate the method.
AB - A quantitative description of dynamical systems requires the estimation of uncertain kinetic parameters and an analysis of their precision. A method frequently used to describe the confidence intervals of estimated parameters is based on the Fisher-Information-Matrix. The application of this traditional method has two important shortcomings: (i) it gives only lower bounds for the variance of a parameter if the solution of the underlying model equations is non-linear in parameters. (ii) The resulting confidence interval is symmetric with respect to the estimated parameter. Here, we show that by applying the bootstrap method a better approximation of (possibly) asymmetric confidence intervals for parameters could be obtained. In contrast to previous applications devoted to non-parametric problems, a dynamical model describing a bio-chemical network is used to evaluate the method.
KW - Bootstrap approach
KW - Fisher-Information-Matrix
KW - Parameter confidence interval
UR - http://www.scopus.com/inward/record.url?scp=33748057502&partnerID=8YFLogxK
U2 - 10.1016/j.ymben.2006.04.003
DO - 10.1016/j.ymben.2006.04.003
M3 - Article
C2 - 16793301
AN - SCOPUS:33748057502
SN - 1096-7176
VL - 8
SP - 447
EP - 455
JO - Metabolic Engineering
JF - Metabolic Engineering
IS - 5
ER -