Decision times and errors increase when classifying trunk postures near posture bin boundaries

Abstract

The purpose of this study was to quantify decision time and classification errors made during trunk posture assessment tasks using a computerised sampling approach. Ninety (45 male, 45 female) university-aged participants, who had varying expertise with body posture evaluation, were asked to accurately classify a series of static images of the trunk using a computer interface by selecting the posture category or ‘bin’, within which the presented posture fell. Images were presented randomly in either a flexion/extension or lateral bending view at various distances from the boundary lines of each posture bin. Subjects with the least experience performed the tasks in the shortest amount of time; however, they also made the most errors. Expert participants reduced the number of classification errors by increasing the time of each posture classification decision. The flexion/extension task proved to take more time and resulted in more errors than the lateral bend task. The closer the presented images were to the bin boundaries, the greater was the time taken and errors made during posture classification. These data have implications for the design of posture classification biomechanical software interfaces and the training of operators who utilise the software.