Physica A 391 (2012) 4933–4942 Contents lists available at SciVerse ScienceDirect Physica A journal homepage: www.elsevier.com/locate/physa On the length of stabilograms: A study performed with detrended fluctuation analysis M. Teresa Blázquez a , Marta Anguiano a , Fernando Arias de Saavedra a , Antonio M. Lallena a,∗ , Pedro Carpena b a Departamento de Física Atómica, Molecular y Nuclear, Universidad de Granada, E-18071 Granada, Spain b Departamento de Física Aplicada II, E.T.S.I. de Telecomunicación, Universidad de Málaga, E-29071 Málaga, Spain article info Article history: Received 21 December 2011 Received in revised form 18 April 2012 Available online 17 May 2012 Keywords: Stabilogram length Human postural control Detrended fluctuation analysis Signal length abstract The effects associated to the length of stabilograms, a measure of the time dependence of the center of pressure of an individual standing up, are analyzed. The fractal characteristics of 27 signals with a length of 2 14 points, each one corresponding to a different individual, are studied by using the Detrended Fluctuation Analysis technique. The properties of the complete signals are compared to those of various subsignals extracted from them. No differences have been found between the characteristic exponents found for x and y signals. The relation between the exponents of the position and velocity signals is accomplished by the 2 14 point signals, while subsignals with up to 2 12 points do not verify it. Using artificial signals with 2 14 points, generated for α values given, it has been demonstrated that the exponents obtained from these signals take values larger than expected for α< 0.3, while the exponents of the accumulated series are smaller than expected for 0.7 <α. For CoP trajectories this indicates that DFA-1 provides feasible exponents for the short τ -end region of the velocity signal and the large τ -end region of the accumulated (position) one. It has been found that the characteristic exponents vary along the series. A slightly larger persistence is found in the last part of the signal for large frequencies in the x direction. © 2012 Elsevier B.V. All rights reserved. 1. Introduction The human postural control system has been a worth topic in the last four decades [1]. It includes the so-called proprioceptive, vestibular and visual systems which are responsible for the human body equilibrium in statical and dynamical conditions. These three systems take advantage of a backup mechanism that ensures an adequate balance of the body if some of them fails [2]. A consequence of this complex procedure is the postural sway. A stabilogram is a measure of this sway, that is, a registration of the variation with time of the center of pressure (CoP) of a subject staying on top of a force platform. This signal is expected to be used to diagnose equilibrium diseases with different origins. Stabilograms have been analyzed from different points of view. From the simple statistical study of the CoP [1] to the search of multifractal behaviors of the stabilograms [3], a variety of different signal analysis techniques have been used. An interesting point related to CoP trajectory characteristics concerns its length. The various analysis performed have considered rather different signal length without paying special attention to it. Just to mention a few examples, sampling times of 15 s [4], 30 s [3,5,6], 32 s [7,8], 40 s [9], 51.2 s [10–12], 90 s [13] and even 30 min [14] have been considered. In addition several sampling frequencies (32, 40, 100 Hz) have been used. As a consequence, the signals analyzed show very ∗ Corresponding author. E-mail addresses: mblazquez@ugr.es (M.T. Blázquez), lallena@ugr.es (A.M. Lallena). 0378-4371/$ – see front matter © 2012 Elsevier B.V. All rights reserved. doi:10.1016/j.physa.2012.05.010