Department of Mechanical Engineering, University of Connecticut, Storrs, CT, USA.
Physiol Meas. 2012 Jan;33(1):79-93. doi: 10.1088/0967-3334/33/1/79.
Non-invasive estimation of minute ventilation is important for quantifying the intensity of physical activity of individuals. In this paper, several improved regression models are presented, based on the measurement of chest and abdomen movements from sensor belts worn by subjects (n = 50) engaged in 14 types of physical activity. Five linear models involving a combination of 11 features were developed, and the effects of different model training approaches and window sizes for computing the features were investigated. The performance of the models was evaluated using experimental data collected during the physical activity protocol. The predicted minute ventilation was compared to the criterion ventilation measured using a bidirectional digital volume transducer housed in a respiratory gas exchange system. The results indicate that the inclusion of breathing frequency and the use of percentile points instead of interdecile ranges over a 60 s window size reduced error by about 43%, when applied to the classical two-degrees-of-freedom model. The mean percentage error of the minute ventilation estimated for all the activities was below 7.5%, verifying reasonably good performance of the models and the applicability of the wearable sensing system for minute ventilation estimation during physical activity.
非侵入式估计分钟通气量对于量化个体的身体活动强度非常重要。在本文中，我们提出了几种改进的回归模型，这些模型基于佩戴在受试者身上的传感器带测量胸部和腹部运动（n=50），涉及 14 种不同的身体活动类型。我们开发了 5 个涉及 11 个特征组合的线性模型，并研究了不同模型训练方法和用于计算特征的窗口大小的影响。使用在身体活动协议期间收集的实验数据评估了模型的性能。将预测的分钟通气量与使用呼吸气体交换系统中内置的双向数字体积换能器测量的标准通气量进行比较。结果表明，当应用于经典的两自由度模型时，包含呼吸频率并使用百分位数点而不是 60 秒窗口大小内的十分位范围，可将误差减少约 43%。所有活动的分钟通气量估计的平均百分比误差均低于 7.5%，这验证了模型具有相当好的性能，以及可穿戴感测系统在身体活动期间估计分钟通气量的适用性。
