I am surprised reading this very interesting paper that the Parkinson patients in the control condition exibited a DFA scaling exponent of 0.9 (Figure 3). Previous studies (Hausdorff et al 2000) found exponent close to 0.5. Moreover, the age of control participant was not matched with patients, and it has been demonstrated that aging changed drastically the DFA scaling exponent. I wonder whether the DFA estimation has been achieved with enough care, in particular if the linear and/or non linear trends have been reduced, as it is well known that those trend (non stationnarity) affect the DFA output (see below).

Tempting long-memory-on the interpretation of DFA results},
author={Maraun, D. and Rust, HW and Timmer, J.},
journal={Nonlinear processes in Geophysics},
volume={11},
number={4},
pages={495--503},
year={2004}

Assessment of long-range correlation in time series: How to avoid pitfalls},
author={Gao, J. and Hu, J. and Tung, W.W. and Cao, Y. and Sarshar, N. and Roychowdhury, V.P.},
journal={Physical Review E},
volume={73},
number={1},
pages={016117},
year={2006},

Integrated approach to the assessment of long range correlation in time series data},
author={Rangarajan, G. and Ding, M.},
journal={Physical Review E},
volume={61},
number={5},
pages={4991},
year={2000},


Dr Julien Lagarde