Ear-Worn Inertial Measurement of Sit-to-Stand and Sit-to-Walk Movements: Algorithm Development and Validation

Sit-to-stand and sit-to-walk are common functional movements that provide insight into an individual's functional mobility. Developing comprehensive measures of these movements using ear-worn inertial measurement units (IMUs), which are increasingly incorporated into widely used devices such as hearing aids and headphones, could provide a means of monitoring older adults and individuals with movement disorders with minimal additional burden.

This paper aimed to introduce comprehensive outcome measures for sit-to-stand and sit-to-walk movements based on data from ear-worn IMUs, and to evaluate their agreement with an optical motion capture reference.

Data were analyzed from 62 participants, of whom 46 (18 male and 28 female; median age 27, IQR 15 years) did not have a known underlying movement condition and 16 (7 male and 9 female; median age 62, IQR 23 years) did. The participants wore a bespoke 3D-printed headset that contained 6-axis IMUs placed beside their left and right ears. A reflective marker was placed on each participant's seventh cervical vertebra to provide an optical motion capture reference. Each participant performed 3 sit-to-stand movements and 3 sit-to-walk movements. The reported outcome metrics included overall and subphase durations, acceleration, velocity, power, movement smoothness, and hesitation. Agreement between ear-derived and optical outcomes was evaluated using linear mixed-effects analyses accounting for repeated trials within participants, including systematic and proportional bias, limits of agreement, root mean squared error, and smallest detectable difference.

Temporal outcomes demonstrated the closest agreement between systems. The mean overall movement duration differed by 0.09 (SD 0.12) seconds for sit-to-stand and 0.06 (SD 0.17) seconds for sit-to-walk, with 95% limits of agreement of -0.14 to 0.32 seconds and -0.28 to 0.39 seconds, respectively. Strong linear relationships were observed for peak power, with mixed-effects regression slopes of 0.70 for sit-to-stand and 0.68 for sit-to-walk, although systematic biases of 172 (95% CI 153-190) W and 191 (95% CI 168-215) W indicated that ear-derived and seventh cervical vertebra-derived power were not interchangeable. Peak upward acceleration similarly showed systematic biases of 1.05 (95% CI 0.93-1.18) and 1.30 ms^-2 (95% CI 1.14-1.45) for sit-to-stand and sit-to-walk, respectively. Hesitation demonstrated comparatively small absolute error (root mean squared error 0.04 and 0.03, respectively), whereas smoothness and other fluency outcomes showed greater disagreement between measurement systems.

The ear-worn IMUs extracted a broad range of sit-to-stand and sit-to-walk outcome measures, although agreement with the optical reference varied between measures. Temporal outcomes demonstrated the closest agreement, whereas kinematic, power, and fluency measures showed varying degrees of systematic and proportional disagreement. These findings support the feasibility of using ear-worn IMUs for comprehensive functional movement assessment and provide a basis for subsequent validation of the most robust outcomes in clinically relevant populations.