Quantification of the validity and reliability of sprint performance metrics computed using inertial sensors: A systematic review

Abstract

Background: Wearable inertial sensors enable sprinting to be biomechanically evaluated in a simple and time efficient manner outside of a laboratory setting.

Research question: Are wearable inertial sensors a valid and reliable method for collecting and measuring sprint performance variables compared to referenced systems?

Methods: PubMed, SPORTDiscus, and Web of Science were searched using the Boolean phrases: ((run* OR sprinting OR sprint*) AND (IMU OR inertial sensor OR wearable sensor OR accelerometer OR gyroscope) AND (valid* OR reliabil*)). Articles with injury-free subjects of any age, sex or activity level were included.

Results: Fifteen studies met the inclusion criteria and were retained for analysis. In summary, higher Intra-class correlation [ICC] or Pearson correlation coefficients (r) were observed for contact time (ICC ≥ 0.80, r ≥ 0.99), trunk angular displacement (r ≥ 0.99), vertical and horizontal force (ICC ≥ 0.88), and theoretical measures of force, velocity and power (r ≥ 0.81). Low coefficient of variation (CV) were found in peak velocity (≤ 1%), average velocity (≤ 3%), and contact time (≤ 3%,). Average and peak velocity, and resultant forces, were found to have a wide range of r (0.32-0.92) and CVs (0.78-20.2%). The lowest r (-0.24 to 0.49) and highest CVs (15-22.4%) were noted for average acceleration, crania-caudal force, instantaneous forces, medio-lateral ground reaction forces, and rate of decrease in ratio of forces.

Significance: Due to a wide range of methodological differences, a clear understanding of the validity and reliability of different inertial sensors for the analysis of sprinting has yet to be established. Future research into the sensor's placement, attachment method and sampling frequency are among several factors that need further investigation.

Keywords: Acceleration; Kinematics; Kinetics; Velocity; Wearable sensors.