Complexity Changes in Human Wrist Temperature Circadian Rhythms through Ageing - Foundations on Natural and Artificial Computation - page 401

generate self-sustained circadian oscillations through positive and negative tran-
scriptional/translational feed-back loops, being Bmal1 , Clock , Per and Cry the
most relevant clock genes [4,12,21].
On the other hand, thermoregulatory processes are essential in mammals. An
homeostatic mechanism keeps the body temperature around a set point, that
is is modulated by circadian rhythmicity. Core temperature decreases during
sleep in order to slow down metabolic activity. This is accomplished by means
of the loss of heat through peripheral blood vessels vasodilation, producing an
increase in skin temperature. Peaks in skin temperature correspond to sleep
stages. The wrist skin temperature, as measured by a wearable temperature
sensor has proven to be a good marker of the circadian rhythms of the subject.
The Chronobiology Laboratory of the University of Murcia designed and vali-
dated a data recording protocol that allows ambulatory assessment of human
circadian rhythms by measuring wrist skin temperature [15,14,10].
Little is known about how circadian system evolves over years. The most
prominent effects would be an amplitude decrease and progressive phase ad-
vances relative to the light dark cycle. Some authors [5] suggest that ageing
is related to instability of circadian phase and increased day-to-day variability.
However, other studies have shown a decrease [6]. Endogenous circadian rhythm
seems to remain stable, but the ability to synchronize the endogenous clock with
the cyclic environment seems to be impaired with age [21]. The term ”robust-
ness” is related to the regularity of circadian rhythms. Formal definition are
lacking although several indexes have been proposed [12].
Physiological complexity is also an elusive but evident feature of living be-
ings. Many biological signals exhibit a complexity loss with age. Complexity
loss is related to structural and functional alterations on systems composed of
many subsystems and feed-back loops [17]. No single comprehensive measure of
biological complexity exists. In practice, complexity measures are based either
in entropy-related concepts or in deterministic nonlinear dynamics. In cardio-
vascular dynamics assessment some useful measures are Approximate Entropy
( ApEn ), detrended fluctuation analysis (DFA) scaling exponent, fractal dimen-
sions or Lyapunov exponents. Circadian studies based on a comprehensive set
of complexity measures are still scarce [19,2].
In this paper we attempt to establish whether complexity changes in hu-
man circadian rhythms in ageing can be assessed through phase irregularities
in individual wrist temperature records. To this end, we analyze what kind of
correlation, if any, exists between age and phase complexity. We propose some
phase complexity measures. A novel approach based on LZ complexity, instan-
taneous phase and Hilbert transform is introduced. The ApEn complexity of
instantaneous phase is also studied, using both the Hilbert transform approach
and a complex continuous wavelet transform (CWT). Our results consistently
show that a significant decrease in phase complexity happens when ageing.
This work is part of a wider research in Ambient Assisted Living frameworks
for the elderly, where detecting changes in the individual circadian pattern could