Potential Usefulness of Tracking Head Movement via a Wearable Device for Equilibrium Function Testing at Home

Abstract

Many studies have reported the use of wearable devices to acquire biological data for the diagnosis and treatment of various diseases. Balance dysfunction, however, is difficult to evaluate in real time because the equilibrium function is conventionally examined using a stabilometer installed on the ground. Here, we used a wearable accelerometer that measures head motion to evaluate balance and examined whether it performs comparably to a conventional stabilometer. We constructed a simplified physical head-feet model that simultaneously records "head" motion measured using an attached wearable accelerometer and center-of-gravity motion at the "feet", which is measured using an attached stabilometer. Total trajectory length (r = 0.818, p -false discovery rate [FDR] = 0.004) and outer peripheral area (r = 0.691, p -FDR = 0.026) values measured using the wearable device and stabilometer were significantly positively correlated. Root mean square area values were not significantly correlated with wearable device stabilometry but were comparable. These results indicate that wearable, widely available, non-medical devices may be used to assess balance outside the hospital setting, and new approaches for testing balance function should be considered.

Keywords: Digital healthcare; Head movement; Postural sway; Stabilometry; Wearable accelerometer sensor.

Fig. 1

Devices used in the experiment. A IMU (Inertial Measurement Unit). B RN002 TW (Wearable devices with accelerometer). C Stabilometer (stabilometric recordings)

Fig. 2

Experiment of head motion measurement data confirmation via stabilometery setup. The IMU and RW002 TW were fixed and attached to the head of the tripod. Data recording continued for 60 s, and a load with a randomly changing position and magnitude was applied to the stabilometer to compare head measurement and stabilometry measurement

Fig. 3

Examination of data by calculating cross-correlation functions. In each figure, the first row represents the displacement data in the IMU, the second row represents the displacement data from the stabilometer or RN002 TW, and the third row represents the cross-correlation function of the data. If the axes are matched, the cross-correlation function signal was observed near the start of the recording and then decayed. If the axes are not matched, the signal was not clear, and noise was high over time

Fig. 4

Correlations between the IMU, RN002 TW, and Stabilometer. A Total length of Trajectory. I_L – length of IMU. R_L – length of RN002 TW. S_L – length of Stabilometer. B Outer peripheral area. I_S – area of IMU. R_S – area of RN002 TW. S_S – area of Stabilometer. C RMS area. I_RMSAREA – area of IMU. R_RMSAREA – area of RN002 TW. S_RMSAREA – area of Stabilometer. The straight line in the graphs represents the regression between the variables. A significant correlation between the total trajectory length and outer peripheral area of head displacement measured via a wearable device and the center-of-gravity movement measured via stabilometry

Fig. 5

Bland–Altman plots along with the mean error and the 95% limits of agreement (CI95%) for comparison between IMU, RN002 TW, and Stabilometer readings. A Total length of Trajectory. B Outer peripheral area. C RMS area