Riddick Ryan, Alshehri Mansour Abdullah, Hodges Paul
School of Health & Rehabilitation Sciences, The University of Queensland, St Lucia, QLD 4072, Australia.
Centre for Innovation in Pain and Health Research (CIPHeR), The University of Queensland, St Lucia, QLD 4072, Australia.
Sensors (Basel). 2025 Aug 13;25(16):5011. doi: 10.3390/s25165011.
Posture, physical activity, and sleep have been shown to be linked to many health issues but are difficult to assess in laboratories, especially in terms of long-term patterns. Worn on the body, inertial measurement units (IMUs) measure motion and have shown promise for longitudinal measurements of these phenomena, but the repeatability of their measurements in daily life has not been extensively characterized. This study assessed the repeatability of measures of spine posture and movement in a set of standardized tasks in the lab versus those performed at home using IMUs. We also evaluated issues that impact data quality for real-world measurements. The results showed moderate repeatability in the range of spinal motion assessed during the tasks (ICC = 0.67). In contrast, the absolute angles of the spine (such as the starting posture) were more variable and more difficult to estimate. The estimation of the reference posture was identified as a key factor. Five methods to estimate the reference posture were compared, and the use of a composite set of standardized tasks performed best (ICC = 0.72 ± 0.17). Additional studies and cross-validation with other sensors are needed to draw stronger conclusions about the optimal methodology. For measurements of daily life over 2 days, magnetic interference had a major impact on the data quality, affecting 43% of all data analyzed. Metrics were developed to assess data quality and strategies are proposed to improve repeatability in future work.
姿势、身体活动和睡眠已被证明与许多健康问题相关，但在实验室中难以评估，尤其是在长期模式方面。惯性测量单元（IMU）佩戴在身体上，可测量运动，并已显示出在纵向测量这些现象方面的潜力，但它们在日常生活中测量的可重复性尚未得到广泛表征。本研究评估了在实验室一组标准化任务中与在家中使用IMU执行任务时脊柱姿势和运动测量的可重复性。我们还评估了影响实际测量数据质量的问题。结果显示，在任务期间评估的脊柱运动范围内，可重复性中等（组内相关系数=0.67）。相比之下，脊柱的绝对角度（如起始姿势）变化更大，更难估计。参考姿势的估计被确定为一个关键因素。比较了五种估计参考姿势的方法，使用一组综合标准化任务的方法表现最佳（组内相关系数=0.72±0.17）。需要进行更多研究并与其他传感器进行交叉验证，以就最佳方法得出更有力的结论。对于超过2天的日常生活测量，磁干扰对数据质量有重大影响，影响了所有分析数据的43%。制定了评估数据质量的指标，并提出了在未来工作中提高可重复性的策略。
