Parallels between homeostatic regulation and control theory

Abstract

Pathologies in which homeostatic regulatory mechanisms are affected tend to impact multiple systems which are connected through an extensive network. Statistical analysis of time series has been explored as a tool to retrieve information about the underlying regulatory dynamics of these networks, to detect subtle changes that could lead to the development of diseases such as diabetes. This approach has been propelled by the introduction of devices capable of recording physiological variables continuously and non invasively. In this study, experimental time series of regulated and regulating physiological variables and an experimental mechatronic control system (a 2-wheeled self-balancing robot) and a corresponding computer model were analyzed. Results showed that in all examined systems variance of the regulating variable adapts in order to keep the regulated variables as constant as possible. Box-and-whisker charts also conveyed information about other important aspects of the regulation, in the physiological case hysteresis and thus non-linearity and memory became apparent, these properties were absent in the other two systems, which showed a linear behavior. Thus, adaptability of the variance of regulated variables seems to be a characteristic that is not only particular to complex regulatory systems and which would be inherent to all feedback control mechanisms. This study suggests that applications from control theory could be useful to better elucidate the connection between time series and the underlying control mechanisms.