Complexity analysis of the temperature curve: new information from body temperature.

An attempt was made to develop a truly quantitative approach to temperature, based on models derived from nonlinear dynamics and chaos theory. Three different procedures for measuring the degree of complexity of the temperature curve were compared, and the possible correlations between these measurements and certain physiopathologically relevant parameters in healthy subjects were examined. Twenty-three healthy subjects (10 males, 13 females) between 18 and 85 years of age had their temperature measured every 10 min for at least 30 h. These time series were used to determine the approximate entropy (ApEn), a detrended fluctuation analysis (DFA), and the fractal dimension by the compass method (FD(c)). There was good correlation between the different methods of measuring the complexity of the curve [ r=-0.603 for ApEn vs. DFA ( p=0.002), r=0.438 for ApEn vs. FDc ( p=0.04) and r=-0.647 for DFA vs. FDc ( p=0.0008)]. Both the fractal dimension and the approximate entropy were inversely correlated with age [ r=-0.637 ( p=0.001) and r=-0.417 ( p=0.03), respectively], while the DFA increased with age ( r=0.413, p=0.04). The results thus suggest that complexity of the temperature curve decreases with age. The complexity of the temperature curve can be quantified in a consistent fashion. Age is associated with lower complexity of the temperature curve.