< Back to Article

Exploration of Circadian Rhythms in Patients with Bilateral Vestibular Loss

Fig 2

Body Temperature curves.

Temperature COSINOR curves (blue) and smoothed curves by local polynomial regression (function “LOESS” in red) for Patients (upper pattern) and Control group (lower pattern). The solid blue lines corresponds to results from the estimated cosinor function The dashed lines around the solid line correspond to a confidence band of level for the estimated cosinor curve. Solid red lines result from local polynomial regression (with span parameter α = 0.25). As before, dashed red lines correspond to the borders of the confidence band of level . Thus, areas in which the two confidence bands do not overlap indicate a difference between the two functions with confidence level 95%. The highest and lowest values estimated by the LOESS method are respectively observed at 4:29 PM and 7:40 AM (although a nearly equally low minimum is already attained much earlier in the morning). The deviations were marginal since the local smoother mostly overlaps the COSINOR modeling only around 5:00 AM and 8:00 AM and around 4:00 PM and 7:00 PM. More precisely, the local smoother indicates that the temperature remains at a lower level than the COSINOR in the morning, with the largest difference attained at about 6:30 AM. In the following hours, the temperature rises faster than the COSINOR captures, and attains its steady state about 12:40 PM. In the following hours, the temperature remains relatively stable at a high level, roughly between 37.25 and 37.35°C. Then, in the evening at about 9:10 PM, the temperature drops, and the decline is stronger than the COSINOR is able to capture.

doi: https://doi.org/10.1371/journal.pone.0155067.g002