Modeling daily and subdaily cycles in rat sleep data.

We developed a statistical model for rat sleep-wake behavior over the 24-hour day; this model could be used for animals exposed to shorter cycles of light and dark conditions. Rat behavior was classified either as "wake" or "sleep," as determined by analysis of electrophysiological data. The proposed model consists of three parts: the first two explain cyclic effects relating to the lighting conditions, whereas the last part reflects any acyclic effect. Hypothesis tests were conducted on the magnitude of the parts. The model also accounts for correlated errors.