Use of inertial measurement units for measuring torso and pelvis orientation, and shoulder–pelvis separation angle in the discus throw

• Published in: International journal of sports science & coaching Vol. 13; no. 6; pp. 985 - 992
• Main Authors: Brice, Sara M, Hurley, Matthew, Phillips, Elissa J
• Format: Journal Article
• Language: English
• Published: London, England SAGE Publications 01-12-2018
• Subjects: track and field athletics; sport biomechanics; Angular kinematics; motion analysis
• ISSN: 1747-9541; 2048-397X
• DOI: 10.1177/1747954118778664

Wearable technologies, such as inertial measurement units, are being increasingly utilised in sport to provide immediate feedback to athletes and coaches on movement dynamics. This study examines the validity of inertial measurement units for measuring data pertinent to discus throwing namely shoulder–pelvis separation angle, and torso and pelvis transverse plane orientation. Five discus throwers performed 10 throws, while shoulder–pelvis separation angle, and torso and pelvis transverse plane orientation were measured simultaneously using a motion capture system and inertial measurement unit system. Time-series torso and pelvis orientation data were compared to determine the validity of the inertial measurement unit system for measuring the segment orientation. Discrete shoulder–pelvis separation angle data were compared to determine the validity of the inertial measurement unit system for measuring the discrete data pertinent to discus throwers and coaches. Discrete data examined were magnitudes of separation that occurred when the torso was maximally rotated to the left and right. Data were compared using root mean square difference and root mean square relative to angle range (RMS%). Bland–Altman analyses were also performed. Torso (RMS% = 3%) and pelvis (RMS% = 2%) orientation data agreed closely. Agreement was lower for separation angle (maximum left rotation RMS% = 9%; maximum right rotation RMS% = 13%). Bland–Altman biases indicate inertial measurement units underestimated segment orientation, underestimated maximum right rotation, and overestimated maximum left rotation. The protocol described was valid for measuring the torso and pelvis orientation. Separation angle validity was low, indicating differences in underlying modelling approaches. Further investigation is needed to examine more optimal sensor positioning, and novel ways of examining shoulder–pelvis dynamics.