Accuracy vs. Practicality of Inertial Measurement Unit Sensors to Evaluate Motor Competence in Children

Abstract

The TGMD (i.e., Test of Gross Motor Development) has been considered as one of the gold standards of assessment tools for analysis of motor competence in children. However, it is rarely used by teachers in schools because the time, resources, and expertise required for one teacher to assess a class of students is prohibitive in most cases. A potential solution may be to automate the testing protocol using objective measures and inertial measurement unit sensors. An accurate method using 17 sensors to capture full body motion profiles and machine learning methods to objectively assess proficiency has been developed; however, feasibility of this method was low. Subsequently, a simplified method using four sensors (i.e., attached to wrists and ankles) was found to be effective, efficient, and potentially highly feasible for use in school settings. For some skills, however, not all skill criteria could be assessed. Additionally, misclassification on occasion, marred results. In the present paper we consider a previous experiment that used wireless motion capture to assess criteria from the TGMD-3. We discuss the advantages alongside the disadvantages of testing motor competence in children using sensors and consider the question—Can a compromise be struck between accuracy and feasibility?