Integrated biosignal data analysis for performance assessment

生物信号为了解人体的基本功能提供了一个窗口。人类的运动包括从大脑传递控制信号来激活骨骼肌，骨骼肌反过来收缩并产生力量和运动。肌肉内的激活信号被检测并记录为肌电图（EMG）。鉴于激活和力之间的联系，从记录的肌电图中预测力的大量工作已经产生。我们已经开发了基于单位点记录的肌电力模型，以及最近基于多位点高密度肌电记录的肌电力模型，其中在研究的肌肉表面上检测到肌肉激活。然而，这些模型仍然受到肌肉激活的不完整图像的限制。同样，我们无法直接测量单个肌肉的输出力（作为替代，测量关节力矩，其中力矩由跨关节的几个肌肉控制）。只有在高度受限的情况下，通常是等距（恒定位置）的实验条件下，才能得到合理的力估计。这在一定程度上是因为肌肉的激活会随着姿势的改变而改变，而姿势的改变会改变系统的生物力学。

Biological signals provide a window into essential functions in the human body. Human movement involves transmission of control signals from the brain to activate skeletal muscles, which in turn, contract and generate force and movement. The activation signal within the muscle is detected and recorded as the electromyogram (EMG). Given the connection between activation and force, an extensive body of work on predicting force from recorded EMG has been generated. We have developed EMG-force models based on single site recordings, and more recently on multi-site high-density EMG recordings, in which muscle activation is detected over a large surface area of the muscle under study. These models, however, are still limited by an incomplete picture of the muscle activation. As well, we are unable to directly measure output forces for individual muscles (as a surrogate, joint moment is measured, where the moment is controlled by several muscles acting across the joint). Reasonable force estimation has been achieved only under highly restricted, usually isometric (constant position), experimental conditions. This is in part because muscle activation changes with postural changes, which alter the biomechanics of the system. Our work has focussed on EMG-based prediction of forces about the elbow joint, in healthy individuals, using advanced signal processing and system modeling techniques. In order to advance our work to develop accurate and reliable force prediction under less restricted, dynamic conditions, we will explicitly include information on the biomechanics of the limb, and on how muscle activation is coordinated, in our force prediction models. The innovative signal processing techniques and model structures that we develop will contribute to our fundamental understanding of normal muscle function and control. In the course of this work, six graduate and four undergraduate students will be trained in the acquisition of EMG data, and the use of advanced processing and modeling techniques. The results of our research will be directly applicable to problems in ergonomics, athletics, and rehabilitation, providing significant benefits to Canada and Canadians.