Fig. 4

Time-dependent complexity characterisation extracted for a subset of the CFS patients using the t-AAA method. Algorithm 3 was applied to the full activity sequence, with a window width of 3 days (72 hours), a step size of 5 minutes and the following settings of the AAA method: \(n_{min} = 1\), \(n_{max} = 9\times 60\), \(s = 1.1\). The solid (blue) curve depicts the evolution of the fractal dimension on a 3-hour scale (calculated using the slope at \(n=3\times 60\)). Note that every point on the curve represents the fractal dimension of the past 3 days. The vertical lines split the recording period into 3 weeks, with the first week only spanning 4 days due to the 3-day window which was applied to obtain the fractal dimensions. For comparison with a static approach, the horizontal dashed lines indicate the weekly fractal dimensions as presented in Table 1. Every week is labelled with the value of the static fractal dimension corresponding to that week, as well as its ranking in terms of functioning. This ranking is also reflected in the background colouring of each week, with red, orange and green corresponding to the worst, average and best week, respectively. Similar figures can be found for the other patients in Fig. 5 in the Appendix