View Full Version : Summary: freeware entropy calculations

David R Collins
04-18-2002, 12:29 AM
Dear Colleagues,

In response to my query regarding freeware software for time series
entropy calculations, I was directed to two pieces of software (from
Frank Borg, borgbros@netti.fi; and zsolt, drchaos@freemail.hu) and an
online package, .
I also received several interesting notes about current projects, more
about which I look forward to learning. Following the messages I
received, I include the original post.

Thank you for your help!

David R. Collins, Ph.D.
Human Performance Laboratory: Barnes-Jewish Hospital
4555 Forest Park Parkway, St. Louis, MO 63108
314-286-1586 (office) 314-286-1568 (fax)

I have recently made a proto softw for posturography that also
calculates the so called approximate entropy, ApEn (Pincus 1991). A
review of its uses and definitions can be found in Nonlinear
Biomedical Signal Processing, vol II, Metin Akay (Ed), Ch 3.

There is a very good free packet of nonlinear time series analysis
software (TISEAN) by Schreiber et al at

with a very good documentation. C-code also listed in the book H Kantz,
T Schreiber: Nonlinear Time Series Analysis, CUP 2000, 1997. In fact I
have implemented and modified their code both w/ C and w/ mathcad. (They
do not discuss the ApEn, however.)

The implementations of the algorithms as such are quite simple. The
hurdle is that computation time may grow drastically with the length of
the data. This is partially remedied by using the so called
neighbourhood searching algorithms which reduces the amount of retracing
of the data (as explained by Screiber et al). I implemented that in the
ApEn algorithm. Anyway, gait and posturogram data are not so demanding
since the sampling rate is quite small and the data size not so big.
ApEn has been popular for studying heart rate. We are about to study
whether it can e.g. discriminate MS patients from *normal* in posturograms.


Frank Borg
Chydenius Institute/ Jyväskylä Univ.
PB 567 Pitkänsillankatu 1-3
FIN-67701 Kokkola Finland
Dear David,
if you still interested in entropy calculations on any movement data I
could give you mine that I used for my Ph.D.
dissertation. However do not expect anything fancy because you vcan only
run it in DOA...but hey it does the job. I
have attached a shor t description of the program (just a basic one) if
you need it let me know and I will tell you the
IP address of my server from where you are going to be able to download it.

bye now:
Hi Dave,

Try TISEAN -- put up by the Kantz & Schreiber group:

Bruce A. Kay, Ph.D., Assistant Professor
Center for the Ecological Study of Perception and Action
Department of Psychology
University of Connecticut
406 Babbidge Rd., Unit 1020, Storrs, CT 06269-1020
Voice: 860-486-4907 Fax: 860-486-2760
Dear Dr. Collins,

You may wish to look at the following publications in which a general
transentropy measure is used to quantify human motion instability. This
measure would be directly applicable to gait evaluation. (If required, I
can send you reprints).

Hatze, H. Motion variability - its definition, quantification and
origin. J. of Motor Behavior 18, 5-16 (1986).

Hatze, H. Perturbation of motions - causes, entropy measures, and
practical applications. In: Högfors, C. Andreasson, G. (edts.), VIII.
Biomechanics Seminar Chalmers University of Technology, Gothenburg,
118-143 (1994).

Hatze, H. The extended transentropy function as a useful
quantifier of human motion variability. Medicine and Science in Sports
and Exercise 27, 751-759 (1995).

Hope this helps,
Best wishes

H. Hatze

Prof. Dr. Herbert Hatze
Head, Department and Laboratory of Biomechanics, ISW
University of Vienna
Auf der Schmelz 6 Tel.: + 43 1 4277 48880
A-1150 Wien Fax: + 43 1 4277 48889
AUSTRIA E-mail: herbert.hatze@univie.ac.at

Please tell me about anything you come up with for this. I have been
thinking about different ways of quantifying the complexity of
trajectories through a VR space, to see if information gathering from
limited movie frames will nudge observers toward specific forms.
Complexity/information measures would be a good one to look at. I hope
you are doing well.

Alex Shull
Cognitive Science and Psychology Programs
Indiana University Bloomington
lab and office: 855-1544
email: jashull@indiana.edu
Hello Mr Collins,

I was very excited to receive a forward of your posting on biomech-l
from a friend. I am also interested in variability issues regarding gait
and the application of dynamic systems to this particular area of study.
I am a doctoral student at the University of Michigan and currently
beginning a research project also involving children with CP and the
effects of Botox injection on the complexity observed. I am using
Lyapunov's exponent to define variability/complexity of the hip and
knee trajectory pre and post injection. These measures will be compared
to the more traditional measures of variabilty (mean and maximum SE of
joint trajectory). It is my hopothesis that the Lyapunov's exponent will
be a more sensitive measure of changes in dynamics that occurs as a
result of the Botox injection... that may not be provided by the more
traditional SE measurements.
I have found John Dingwells paper from the Journal of Biomechanical
Engineering, Feb. 2001 to be particularly helpful, as well, the Journal
of Applied Biomechanics published a number of articles regarding
variability and gait that may be of interest in 2000, vol 16(4). I hope
these may be of some use.. assuming you haven't already found them!

I would love to hear of any research projects or insights that you
may have regarding variability and gait.

Take care,

Sandra McKay
Dear David,
I am not absolutely sure about the entropy measures that you are referring
to, but I would like to point out KineView for you. The software was
initially intended for gait analysis and therefore it provides you with
tools that will enable you to take measurements for gait. You can look
it up on our website at www.kine.is and you can also order a demo if you
would like.
I hope this can be of some help to you.

Carola Frank A?albjornsson, Ph.D.
Research & Development
Kine ehf. Skulagata 26
101 Reykjavik Iceland
Phone # (354) 580-8308/ 580-8300
fax# (354) 580-8309
Original Post:
Dear colleagues,

I could use some help finding programs or code to perform entropy
(information) calculations on gait data, along with any
recommendations for which of the many particular versions of entropy
is best suited for movement data. Ready-made freeware programs for
Windows or Dos, or code in C/C++ or Matlab would be ideal, but I
could adapt Pascal or Fortran code fairly readily if the algorithms
are straightforward. (There are many sophisticated programs with time
series methods, but at $400 to $1000+ they are not cheap, nor as
flexible as Matlab. Also, I would like to know at least conceptually
just how things are being calculated, without undocumented
proprietary tricks.)

The goal of the entropy measure is to determine whether patients with
cerebral palsy exhibit more or less complex dynamics before and after
a surgical intervention, and how the pre/post data compares to
able-bodied data. The data is from rather irregular gait, so no
direct time series comparisons are obvious, and there is no standard
against which to calculate standard deviations. I have gotten
together a fair literature search and stack of articles, and have
been busy catching up on the variety of entropy (/information/
complexity/ predictability) measures from nonlinear dynamics, physics
and information theory. A measure that is related to phase space
reconstruction would be great so that I can examine dimensionality

The only freeware program that calculates entropy that I am aware of
so far is the Recurrence Quantification Analysis software from Webber
and Zbilut:
The Biomch-L archives only include references to conferences, e.g.

Thank you for you help!

David R. Collins, Ph.D.
Human Performance Laboratory: Barnes-Jewish Hospital
4555 Forest Park Parkway, St. Louis, MO 63108
314-286-1586 (office) 314-286-1568 (fax)

To unsubscribe send SIGNOFF BIOMCH-L to LISTSERV@nic.surfnet.nl
For information and archives: http://isb.ri.ccf.org/biomch-l