Share Email Print

Proceedings Paper

Confidence intervals for nonlinear regression procedures
Author(s): Michael L. Johnson; Martin Straume
Format Member Price Non-Member Price
PDF $17.00 $21.00

Paper Abstract

There is no closed form method for the evaluation of the precision of parameter estimates for nonlinear regression problems. There are, however, a number of approximate methods which require different restrictive assumptions and require different amounts of computer time. While none of these estimates are exactly correct, some provide substantially better results than others. This work discusses some of these methods with specific application to frequency domain fluorescence lifetime measurements. A significant conclusion being that the commonly used asymptotic standard errors do not provide reasonable estimates of the precision of parameter values for nonlinear regression problems.

Paper Details

Date Published: 1 April 1996
PDF: 8 pages
Proc. SPIE 2680, Ultrasensitive Biochemical Diagnostics, (1 April 1996); doi: 10.1117/12.237639
Show Author Affiliations
Michael L. Johnson, Univ. of Virginia Health Sciences Ctr. (United States)
Martin Straume, Univ. of Virginia Health Sciences Ctr. (United States)

Published in SPIE Proceedings Vol. 2680:
Ultrasensitive Biochemical Diagnostics
Gerald E. Cohn; Steven A. Soper; C. H. Winston Chen, Editor(s)

© SPIE. Terms of Use
Back to Top