(J-400) Balance Assessment by a Handheld Smartphone with Principal Component Analysis for Anatomical Calibration

Introduction:: Inertial measurement units (IMUs) are becoming an increasingly promising solution for balance assessment [1] and have opened new opportunities for evaluating balance as a treatment response indicator for patients with neurological ailments like Parkinson’s disease, epilepsy, and multiple sclerosis [2]. IMUs, including those found in smartphones, have proven to be a reliable tool in analyzing static balance poses compared to force plates, which have generally been considered the gold standard in measuring balance and postural stability [3]. However, most balance studies using smartphones have strictly placed them with a body strap while assuming alignment of the smartphone and anatomical body axes. Such strict placement may introduce errors (e.g., due to natural curvatures of the spine if the smartphone is placed on the lower back). To address that limitation, in this study we used an anatomical alignment method, proposed in a recent balance study using smartwatches [4], so that the smartphone can be held by the participant in a comfortable position. This method utilizes a single calibration maneuver and Principal Component Analysis (PCA) to align the smartphone and anatomical body axes. The objectives of this study were to (1) determine if there exist correlations between angular velocity scores from a handheld smartphone with PCA alignment and a strictly placed smartphone during a forward flexion calibration maneuver and (2) analyze within-participant acceleration score differences across three balance poses of increasing difficulty for both the handheld and strictly placed smartphones.

Materials and Methods:: 10 subjects (3 females, 7 males, aged 19-21) who provided informed consent were fitted with two iPhone 11’s (Apple, CA, USA): one in a handheld position at the shoulder (Figure 1) and one in a body strap on the lower back. Before each pose, a forward flexion calibration maneuver was performed and its angular velocity data were used in a PCA axis alignment algorithm to identify the anterior-posterior (AP), medial-lateral (ML), and superior-inferior (SI) axes [4]. Participants performed three balance poses for 30 seconds each: (1) double-leg stance with both feet on ground at shoulder width, (2) tandem stance in a heel-to-toe position with the dominant foot leading, and (3) single-leg stance with the non-dominant foot on ground and the dominant leg bent 90 degrees. After each pose, a jump was performed to mark the end of data collection. Both smartphones recorded inertial acceleration, angular velocity, and angular orientation at 100 Hz using the MATLAB App (MathWorks, MA, USA). A 3.5 Hz cutoff, zero-phase, 4th order Butterworth filter was applied to angular velocity and acceleration data. Post-processing and PCA algorithms were performed using MATLAB to reduce the data to Root Mean Square (RMS) angular velocity and acceleration scores in AP, ML, and SI directions. RMS angular velocity scores for both smartphones in the calibration maneuver were analyzed with Pearson correlations. Repeated-measures ANOVA (RMANOVA) was performed to measure within-participant differences in RMS acceleration scores for poses of increasing difficulty.

Results, Conclusions, and Discussions::
With the increasing power of smartphone IMUs, their potential use as measurement tools in balance assessments will only become more prolific for both physicians and patients. The results of this study strongly suggest that 1) it is viable to use a handheld smartphone with a single calibration maneuver and PCA to align the smartphone and anatomical axes for balance assessment and 2) these novel methods allow for a smartphone to be held in a comfortable position for balance tests and, thus, make “at home” balance assessments more accessible and accurate for anyone who has access to a smartphone. Several limitations of this study include its low sample size, relatively narrow age range, and lack of validation with force plate data – which is considered the gold standard method for balance assessment. Regardless of these limitations, the results suggest that the novel methods developed here are ready to be used with smartphones to better characterize differences in balance ability among populations that are at risk of poor balance and postural stability.

Acknowledgements (Optional): : Funding was provided by the Donald E. Bently Center for Engineering Innovation (SK).

References (Optional): : [1] Felius et al., Gait & Posture, 98:62-68, 2022. [2] Bernhard et al., BMC Neurology, 18(1): 114, 2018. [3] Hansen et al., International Journal of Environmental Research and Public Health, 18(7): 3644, 2021. [4] Presley et al., Sensors, 23(10): 4585, 2023.