![Sample our Computer Science journals, sign in here to start your access, latest two full volumes FREE to you for 14 days]({"type":"image" , "src" : "/sda/5928/TOC-Subject-Sample-2021-Computer-Science.jpg"})
Abstract
The purpose of this study was to examine the influence of training on the decision times and errors associated with video-based trunk posture classifications. Altogether, 30 amateur and 30 knowledge-based participants completed a three-phase study (pre-training, training, post-training) that required them to classify static trunk postures in images on a computer screen into a posture category that represented the angle of the trunk depicted. Trunk postures were presented in both flexion/extension and lateral bend views and at several distances from the boundaries of the posture categories. Both decision time and errors decreased as distance from the boundaries increased. On average, amateur analysts experienced a larger decrease in decision time per posture classification than knowledge-based analysts (amateur: 0.79 s, knowledge-based: 0.60 s; p <0.05) suggesting that training can have beneficial effects on classification performance. The implications are that the analysis time associated with video-based posture assessment methods can be reduced with appropriate training, making this type of approach feasible for larger-scale field studies.
Statement of Relevance:Understanding the role that training can play in reducing errors associated with the use of video-based posture assessment methods may result in more efficient use of these tools by ergonomic practitioners. Reducing decision time and misclassification errors will provide a more efficient, accurate and representative assessment of injury risk.
Acknowledgements
This study was funded by the AUTO21 Network Centres of Excellence, whose funding is provided by the Canadian Federal Government. Dr Jack P. Callaghan is supported by a Canada Research Chair in Spine Biomechanics and Injury Prevention.