Autoregressive growth curves and Kalman filtering
- 1 January 1988
- journal article
- methods for-analysing-repeated-measurements
- Published by Wiley in Statistics in Medicine
- Vol. 7 (1-2), 73-86
- https://doi.org/10.1002/sim.4780070111
Abstract
The first part of this paper describes how a Kalman filter can be used to construct maximum likelihood (ML) estimates of autoregressive (AR) and polynomial parameters in polynomial growth curves with AR-1 errors and irregularly-spaced data. The second part introduces a disturbed highest derivative polynomial (DHDP) as a model for growth curves. This model does not depend on regression coefficients. Variances of the highest derivative disturbance and the observation error are estimated (by ML) using a Kalman filter. The estimated DHDP growth curve is obtained by optimally smoothing the output of the filter. Equally spaced data is not required. The DHDP model and analysis are developed for an individual and extended to a population growth curve using data from many individuals with covariates.Keywords
This publication has 18 references indexed in Scilit:
- Time series regression with unequally spaced dataJournal of Applied Probability, 1986
- Linear Models for the Analysis of Longitudinal StudiesThe American Statistician, 1985
- The Signal Extraction Approach to Nonlinear Regression and Spline SmoothingJournal of the American Statistical Association, 1983
- Tracking: Prediction of Future Values from Serial MeasurementsPublished by JSTOR ,1981
- Maximum Likelihood Fitting of ARMA Models to Time Series With Missing ObservationsTechnometrics, 1980
- Testing a Nonlinear Regression Specification: A Nonregular CaseJournal of the American Statistical Association, 1977
- Maximum Likelihood Approaches to Variance Component Estimation and to Related ProblemsJournal of the American Statistical Association, 1977