Automated detection and explainability of pathological gait patterns using a one-class support vector machine trained on inertial measurement unit based gait data.

Journal: Clinical biomechanics (Bristol, Avon)
Published Date:

Abstract

BACKGROUND: Machine learning approaches for the classification of pathological gait based on kinematic data, e.g. derived from inertial sensors, are commonly used in terms of a multi-class classification problem. However, there is a lack of research regarding one-class classifiers that are independent of certain pathologies. Therefore, it was the aim of this work to design a one-class classifier based on healthy norm-data that provides not only a prediction probability but rather an explanation of the classification decision, increasing the acceptance of this machine learning approach.

Authors

  • Wolfgang Teufl
    Junior Research Group wearHEALTH, Technische Universität Kaiserslautern, Gottlieb-Daimler-Str. 48, 67663 Kaiserslautern, Germany.
  • Bertram Taetz
    Junior Research Group wearHEALTH, University of Kaiserslautern, Gottlieb-Daimler-Str. 48, 67663 Kaiserslautern, Germany. taetz@cs.uni-kl.de.
  • Markus Miezal
    Technische Universität Kaiserslautern, Department of Computer Science, Gottlieb-Daimler-Straße 48, 67663 Kaiserslautern, Germany. Electronic address: miezal@cs.uni-kl.de.
  • Carlo Dindorf
    Department of Sports Science, Technische Universität Kaiserslautern, Erwin-Schrödinger-Str. 57, 67663 Kaiserslautern, Germany.
  • Michael Fröhlich
    Department of Sports Science, Technische Universität Kaiserslautern, Erwin-Schrödinger-Str. 57, 67663 Kaiserslautern, Germany.
  • Ursula Trinler
    BG Klinik Ludwigshafen, Ludwig-Guttmann-Straße 13, 67071 Ludwigshafen, Germany. Electronic address: ursula.trinler@bgu-ludwigshafen.de.
  • Aidan Hogan
    Centre for Semantic Web Research, DCC, University of Chile, Santiago, Chile.
  • Gabriele Bleser
    Junior Research Group wearHEALTH, University of Kaiserslautern, Gottlieb-Daimler-Str. 48, 67663 Kaiserslautern, Germany. bleser@cs.uni-kl.de.