Neural network based time zone classification allows circadian rhythm analysis

The circadian rhythm modulates memory in animals. However, the relationship between the circadian cycle and brain wave dynamics has not yet been clarified. Beta wave-like oscillations with frequencies in the range of 13–30 Hz can be induced by applying a cholinergic agonist, carbachol (CCh). We estimated the time of day using a neural network with oscillatory parameters. Rats were kept under light/dark (LD) condition or dark/dark (DD) condition, and hippocampal slices were used for analysis. CCh-induced beta oscillation (CIBO) was induced with CCh at all time zones in the four-hour bins. CIBO showed significant differences in duration, and inter-burst interval (IBI). Additionally, there were significant differences between LD and DD conditions within the same time zones. CIBO duration, IBI and amplitude were not affected by GABA
A
receptor antagonist (SR95531) during the light phase, which is different from our previous results on CIBO frequency. Finally, neural network classifiers that estimated six time zones using CIBO parameters were developed and evaluated. We obtained an average of accuracy rate of 78%. Our results suggest that the brain wave changes depend on the time of day, and the inhibitory neural system may be involved in the change partially. Further, we showed that it is possible to develop time zone classifiers using brain waves.