Evaluation of three methods for determining EMG-muscle force parameter estimates for the shoulder muscles

Christopher J. Gatti, Lisa Case Doro, Joseph E. Langenderfer, Amy G. Mell, Joseph D. Maratt, James E. Carpenter, Richard E. Hughes

Research output: Contribution to journalArticlepeer-review

23 Scopus citations

Abstract

Background: Accurate prediction of in vivo muscle forces is essential for relevant analyses of musculoskeletal biomechanics. The purpose of this study was to evaluate three methods for predicting muscle forces of the shoulder by comparing calculated muscle parameters, which relate electromyographic activity to muscle forces. Methods: Thirteen subjects performed sub-maximal, isometric contractions consisting of six actions about the shoulder and two actions about the elbow. Electromyography from 12 shoulder muscles and internal shoulder moments were used to determine muscle parameters using traditional multiple linear regression, principal-components regression, and a sequential muscle parameter determination process using principal-components regression. Muscle parameters were evaluated based on their sign (positive or negative), standard deviations, and error between the measured and predicted internal shoulder moments. Findings: It was found that no method was superior with respect to all evaluation criteria. The sequential principal-components regression method most frequently produced muscle parameters that could be used to estimate muscle forces, multiple regression best predicted the measured internal shoulder moments, and the results of principal-components regression fell between those of sequential principal-components regression and multiple regression. Interpretation: The selection of a muscle parameter estimation method should be based on the importance of the evaluation criteria. Sequential principal-components regression should be used if a greater number of physiologically accurate muscle forces are desired, while multiple regression should be used for a more accurate prediction of measured internal shoulder moments. However, all methods produced muscle parameters which can be used to predict in vivo muscle forces of the shoulder.

Original languageEnglish
Pages (from-to)166-174
Number of pages9
JournalClinical Biomechanics
Volume23
Issue number2
DOIs
StatePublished - Feb 2008

Keywords

  • EMG
  • Muscle force model
  • Principal-components regression
  • Shoulder

Fingerprint

Dive into the research topics of 'Evaluation of three methods for determining EMG-muscle force parameter estimates for the shoulder muscles'. Together they form a unique fingerprint.

Cite this