The study of biological rhythms is receiving a lot of attention in the literature in
recent years. At the core of this research lies the methodological problem of how to
detect rhythmic signals in measured data. Night and day, or dark and light patterns
impact on human health in many different ways. For this reason, researchers are
studying the effect of sleep on the circadian clock in human body during various stages
of life. Important components of this clock are the circadian genes which have rhythmic
expression overtime with phases suitably matching the night and day. Consequently,
the identification of rhythmic signals is a problem of considerable interest for biologists.
In this work, we develop a novel statistical procedure to detect rhythmic signals in
oscillatory systems based on Order Restricted Inference (ORI). This methodology is
tested both on simulations and on real data bases. Moreover the obtained results are
compared with the most widely extended rhythmicity detection algorithms in literature.