Evaluating reliability of hidden Markov models that describe the lifting patterns of chronic lower back pain patients and controls

Conf Proc IEEE Eng Med Biol Soc. 2006:2006:3238-41. doi: 10.1109/IEMBS.2006.260617.

Abstract

Two hidden Markov models (HMMs) were designed to identify sub-groups of chronic lower back pain (CLBP) subjects based on time series of lifting parameters obtained during a repetitive lifting task. Two simulation studies were conducted to determine the reliability of this approach, using data from the repetitive lifting study. The first simulation verifies that control and CLBP HMMs based on these data can reliably identify sequences that were generated from that model. The second simulation determines whether the HMMs can reliably identify sequences that are intentionally misclassified (CLBP lifting sequences included in the control group and visa versa). The kappa statistic is used to quantify reliability. The simulation results show that the HMMs provide a reliable technique to analyze time series of lifting patterns and can be used to identify misclassified subjects as a subgroup.

Publication types

  • Evaluation Study
  • Research Support, N.I.H., Extramural

MeSH terms

  • Adult
  • Aged
  • Biomechanical Phenomena
  • Biomedical Engineering
  • Case-Control Studies
  • Computer Simulation
  • Humans
  • Lifting*
  • Low Back Pain / classification
  • Low Back Pain / physiopathology*
  • Markov Chains*
  • Middle Aged
  • Models, Biological*