Extracting interpretable muscle activation patterns with time series knowledge mining

Article type: Research Article

Authors: Mörchen, Fabian^{a; *} | Ultsch, Alfred^a | Hoos, Olaf^b

Affiliations: [a] Data Bionics Research Group, University of Marburg, Hans-Meerwein-Str., D-35032 Marburg, Germany | [b] Department of Sports Medicine, University of Marburg, D-35032 Marburg, Germany

Correspondence: [*] Corresponding author. Tel.: +49 6421 2823922; Fax: +49 6421 2828902; E-mail: fabian@informatik.uni-marburg.de

Abstract: The understanding of complex muscle coordination is an important goal in human movement science. There are numerous applications in medicine, sports, and robotics. The coordination process can be studied by observing complex, often cyclic movements, which are dynamically repeated in an almost identical manner. The muscle activation is measured using kinesiological EMG. Mining the EMG data to identify patterns, which explain the interplay and coordination of muscles is a very difficult Knowledge Discovery task. We present the Time Series Knowledge Mining framework to discover knowledge in multivariate time series and show how it can be used to extract such temporal patterns.

DOI: 10.3233/KES-2005-9304

Journal: International Journal of Knowledge-based and Intelligent Engineering Systems, vol. 9, no. 3, pp. 197-208, 2005