Abstract
The methodology for analyzing a biological rhythm has already been the subject of much investigation. However, many questions still have no answer, for example, questions such as the periods chosen: fixed or determined. Throughout this article, we suggest a somewhat innovative methodology that makes it possible to define the steps that seem essential to us in the scientific analysis of rhythms. For some time, this methodology has been put into practice in our laboratory in various studies, some of which have given rise to publications. The notion of quality is a new notion that is present in industry and, when applied to sampling, can improve experimentation. In this way, one may judge the degree to which data samples can be explored as well as the degree of validity of the results of exploration. We provide several methods for achieving this. The search for periods is also important. For this, we have various methods but we must be able to determine those that are the most appropriate and reliable in a particular case. We propose spectral methods, two of which are new and complement ‘Cosinor’ methodology. On the other hand, modelling uses various methods such as those from, for example, periodic trigonometric functions or more complex chaos functions. We are interested in models from the field of regression (the cosine model) and complementary statistical tests that make it possible to validate the proposed model.
Acknowledgement
The graphs that come from the Time Series Analysis Seriel Cosinor software are from the laboratory of Applied Statistics (http://www.euroestech.com). The various samples of data come from studies that we have carried out and that have given rise to publication or are in preparation for publication.