Statistical Models for Establishing a Control Data set for Biomechanical Gait Analysis

这个提案形成了皇后大学（Dr. K.J. Deluzio）和一家加拿大公司（HAS-Motion, Kingston， ON）之间的新合作。目的是为人体步态的三维运动捕获数据制定质量保证标准。HAS-Motion的主要产品Visual3D有助于对一个或少数受试者的生物力学数据进行分析，并输出评估报告。has - motion已经确定需要通过统计处理来增强Visual3D，以促进公共数据共享和多中心研究。此次合作的长期目标是建立一个公共数据库，用于人类步态分析的数据共享。我们最初的里程碑是开发用于评估数据质量和生物力学/功能分类的统计标准。该计划是收集一组健康的年轻受试者在行走、跑步和模拟异常行走时的运动数据。然后，我们将开发一个统计模式识别工具来分类步态模式并预测群体成员（正常步态，跑步步态，扰动步态）。这项研究的结果将对公司有直接的好处，它也直接关系到Deluzio博士的目标，即检测和治疗肌肉骨骼问题。该项目将提升加拿大在生物力学领域和科学领域的地位，作为促进公众获取数据的领导者。需要大量对照数据集的实验室（如临床和军事研究实验室）往往不能将其临床研究资源用于开发适当匹配的对照数据集。从长远来看，拥有足够大的控制和受损条件数据集，该研究可以提高预测和优化患者特异性临床结果的能力，并具有高度的信心。

This proposal forms a new collaboration between Queen's University (Dr. K.J. Deluzio) and a Canadian company (HAS-Motion, Kingston, ON). The aim is to develop quality assurance criteria for 3D motion capture data of human gait. HAS-Motion's primary product, Visual3D, facilitates the analysis of biomechanical data for one, or a small number of subjects, with the output being an assessment report. HAS-Motion has identified a need to enhance Visual3D with statistical processing to facilitate public data sharing and multicenter studies. The long term goal of this collaboration is to build a public database for data sharing in human gait analysis. Our initial milestone is the development of statistical criteria for assessing data quality and biomechanical/functional classification. The plan is to collect motion data from a group of healthy young subjects as they walk, run, and perform simulated abnormal walking. Then we will develop a statistical pattern recognition tool to classify the gait patterns and predict group membership (normal gait, running, perturbed gait). This outcome of this research will have a direct benefit to the company, and it is also directly related to Dr. Deluzio's goal to detect and treat musculoskeletal problems. The project will advance Canada's position within the field of biomechanics and within science in general as a leader in promoting public access to data. Labs requiring large sets of control data (e.g. clinical and military research labs) often cannot devote their clinical research resources to develop appropriately matched control data sets. In the long term, with large enough datasets for control and impaired conditions, the research can improve the ability to predict and optimize patient-specific clinical outcomes with a high level of confidence.