School of Biomedical Engineering, McMaster University, Hamilton, Ontario, Canada.
Department of Kinesiology, McMaster University, Hamilton, Ontario, Canada.
PLoS One. 2024 May 13;19(5):e0290912. doi: 10.1371/journal.pone.0290912. eCollection 2024.
This cross-sectional study aimed to identify and validate cut-points for measuring physical activity using Axivity AX6 accelerometers positioned at the shank in older adults. Free-living physical activity was assessed in 35 adults aged 55 and older, where each participant wore a shank-mounted Axivity and a waist-mounted ActiGraph simultaneously for 72 hours. Optimized cut-points for each participant's Axivity data were determined using an optimization algorithm to align with ActiGraph results. To assess the validity between the physical activity assessments from the optimized Axivity cut-points, a leave-one-out cross-validation was conducted. Bland-Altman plots with 95% limits of agreement, intraclass correlation coefficients (ICC), and mean differences were used for comparing the systems. The results indicated good agreement between the two accelerometers when classifying sedentary behaviour (ICC = 0.85) and light physical activity (ICC = 0.80), and moderate agreement when classifying moderate physical activity (ICC = 0.67) and vigorous physical activity (ICC = 0.70). Upon removal of a significant outlier, the agreement was slightly improved for sedentary behaviour (ICC = 0.86) and light physical activity (ICC = 0.82), but substantially improved for moderate physical activity (ICC = 0.81) and vigorous physical activity (ICC = 0.96). Overall, the study successfully demonstrated the capability of the resultant cut-point model to accurately classify physical activity using Axivity AX6 sensors placed at the shank.
本横断面研究旨在确定并验证使用安信可（Axivity）AX6 腿部加速度计测量老年人身体活动的切点，并对其进行验证。通过让 35 名年龄在 55 岁及以上的成年人佩戴腿部加速度计和腰部加速度计，分别连续 72 小时记录自由活动期间的身体活动，评估其自由活动的身体活动。使用优化算法为每位参与者的 Axivity 数据确定最佳切点，以与 ActiGraph 结果相匹配。为了评估优化后的 Axivity 切点的身体活动评估之间的有效性，进行了一次留一法交叉验证。使用 Bland-Altman 图和 95%一致性界限、组内相关系数（ICC）和平均差异来比较这两个系统。结果表明，当对久坐行为（ICC = 0.85）和低强度身体活动（ICC = 0.80）进行分类时，两种加速度计之间具有良好的一致性，而当对中强度身体活动（ICC = 0.67）和高强度身体活动（ICC = 0.70）进行分类时，一致性为中度。当去除一个显著的离群值后，对久坐行为（ICC = 0.86）和低强度身体活动（ICC = 0.82）的一致性略有提高，但对中强度身体活动（ICC = 0.81）和高强度身体活动（ICC = 0.96）的一致性显著提高。总的来说，该研究成功地证明了利用放置在腿部的安信可（Axivity）AX6 传感器的切点模型来准确分类身体活动的能力。
