Analysis of correlated series of repeated measurements: application to challenge data

• Published in: Biometrical journal Vol. 45; no. 7; pp. 866 - 886
• Main Authors: Engel, Bas, van Reenen, Kees, Buist, Willem
• Format: Journal Article
• Language: English
• Published: Berlin WILEY-VCH Verlag 01-10-2003; WILEY‐VCH Verlag; Wiley-VCH
• Subjects: Approximations and expansions; curves; Eigenfunctions; Exact sciences and technology; growth; inference; Inference from stochastic processes; time series analysis; information; Linear inference, regression; Longitudinal data; Mathematical analysis; Mathematics; maximum-likelihood; mixed effects models; Non-linear mixed models; parameter; Polynomials; Probability and statistics; Random coefficients; REML; Sciences and techniques of general use; smoothing spline; Splines; Statistics; variance; B spline; Integration; Eigenfunction; Time series; Statistical estimation; Parametric model; Stochastic process; Laplacian; Spline; Semiparametric model; Parametric method; Correlation analysis; Population curve; Mixed model; Non linear model; Individual curve; Repeated measurement; Random effect; Software; Maximum likelihood; Autocorrelation; Principal component analysis
• DOI: 10.1002/bimj.200390055
• EISSN: 1521-4036
• ISSN: 0323-3847

Data from an ACTH challenge experiment with veal calves, two short time series of cortisol measurements per animal at 6 and 27 weeks of age, are analysed. Interest is focussed on variation in cortisol profiles both within and between animals. Potential effects of an animals diet and housing system on the profiles are addressed as well. Fully parametric and semi‐parametric models, combining individual random effects with effects of diet and housing, were fitted using (approximate) restricted maximum likelihood (employing Laplacian integration). Eigenfunctions were utilized to describe the variation between profiles and the connection between profiles of the same individual. All calculations were performed with standard software. Results of the analysis provides empirical support for the existence of stable individual characteristics mediating reactivity of the adrenal cortex.