RE: [NMusers] Modeling Growth Hormone Pulsatility

Hi, Jurgen,

Thanks for your reply and this reference.

Fourier analysis is powerful but for non-stationary and non-linear data the basic assumptions of Fourier analysis are no longer valid. Therefore, I am looking for a method to describe both baseline circadian and pulsatility. Any further suggestions are appreciated.

Jian

Jurgen Bulitta <jbulitta_at_buffalo.edu> wrote: Dear Jian,

A Fourier series with maybe 2-6 SIN and COS functions (and a baseline) should be able to describe multiple peaks. I do not know the shape of the pulses in your systems. Fourier series are very powerful to re-capture periodic functions empirically. If you estimate the coefficients of the SIN and COS functions as fixed effects, you should be fine in terms of overparameterization.

One nice feature of the individual functions in a Fourier series is that these functions are orthogonal towards each other. Therefore, I would expect that the coefficients of the SIN and COS functions do not change (much), if you use 2, 4, or 6, etc. SIN/COS terms.

One reference with an application:
Krzyzanski W, Chakraborty A, Jusko WJ; Algorithm for application of Fourier analysis for biorhythmic baselines of pharmacodynamic indirect response models. Chronobiol Int; 2000; 17 77-93

Best regards
Juergen


-----------------------------------------------
Juergen Bulitta, PhD, Post-doctoral Fellow
Pharmacometrics, University at Buffalo, NY, USA
Phone: +1 716 645 2855 ext. 281, j_at_bulitta.com
-----------------------------------------------




 
---------------------------------
We won't tell. Get more on shows you hate to love
(and love to hate): Yahoo! TV's Guilty Pleasures list.
Received on Sat Mar 31 2007 - 00:17:52 EDT