Rapid energy expenditure estimation for assisted and inclined loaded walking

This project aims to estimate energy expenditure during ankle assisted and inclined loaded walking from wearable sensor signals. In this work we trained models using input features of muscle activity (EMG) and ground reaction forces to estimate energy expenditure during walking (1) with an ankle exoskeleton and (2) during various loaded and inclined walking conditions. These estimates were made every single gait cycle. We measure the performance of these models for cases where conditions or subjects data are not included in the training data to simulate real use cases. For research or clinical experiments where the accuracy of our models is sufficient these can be used to replace the standard indirect calorimetry methods of estimating energy expenditure which require expensive equipment and provide slow, noisy measurements. This repository contains the relevant formatted datasets, code to train models, and final model parameters to recreate our reported results. This code base should offer a support to apply our techniques to new datasets or utilize our fully trained models. <br/><br/>This project includes the following software/data packages: <br/> <ul> <li> <a href="https://simtk.org/frs?group_id=1797#pack_2155">EnergyExpenditureEstimation </a> : We encourage the replication and extension of this project. Here we provide links to the biomechanics datasets, code to train learning models, and final model parameters for our work in rapid energy expenditure estimation for assisted and inclined loaded walking. </li> </ul>

本课题旨在通过穿戴式传感器信号，对踝关节辅助及倾斜负载行走过程中的能量消耗进行估算。本研究中，我们采用肌肉活动（EMG）和地面反作用力作为输入特征，训练模型以估算行走过程中的能量消耗，包括：（1）使用踝关节外骨骼；（2）在多种负载和倾斜行走条件下。这些估算在每个步态周期内进行。我们测量了模型在条件或受试者数据未包含在训练数据中的情况下的性能，以模拟实际应用场景。对于研究或临床实验，若我们模型的准确性足够，则可使用这些模型替代标准间接量热法估算能量消耗，后者需要昂贵的设备且提供测量缓慢、噪声较大的结果。本仓库包含相关的格式化数据集、训练模型的代码以及再现我们报道结果的最终模型参数。此代码库应能支持将我们的技术应用于新的数据集或使用我们的完全训练好的模型。

本项目包括以下软件/数据包：

<ul>
<li><a href="https://simtk.org/frs?group_id=1797#pack_2155">能量消耗估算</a>：我们鼓励对该项目的复制和扩展。在此，我们提供了生物力学数据集链接、训练学习模型的代码以及我们快速估算辅助和倾斜负载行走能量消耗工作的最终模型参数。</li>
</ul>