**3. Results**

Representative time series data for walking with head turns L/R and walking with head turns U/D are displayed for one healthy participant (Figure 1) and one participant with mTBI (Figure 2).

**Figure 1.** Representative time series data from one healthy participant. Each subplot provides the optical signal (black solid line) and IMU signal (grey dashed line) overlaid in the upper figure, and the error between optical and IMU signal below (grey solid line). Subplots represent: (**A**) head orientation for walking with head turns (L/R); (**B**) peak rotational velocity of the head for walking with head turns (L/R); (**C**) head orientation for walking with head turns (U/D); and (**D**) peak rotational velocity of the head for walking with head turns (U/D).

**Figure 2.** Representative time series data from one participant with mTBI. Each subplot provides the optical signal (black solid line) and IMU signal (grey dashed line) overlaid in the upper figure, and the error between optical and IMU signal below (grey solid line). Subplots represent: (**A**) head orientation for walking with head turns (L/R); (**B**) peak rotational velocity of the head for walking with head turns (L/R); (**C**) head orientation for walking with head turns (U/D); and (**D**) peak rotational velocity of the head for walking with head turns (U/D).

Correlation, RMSE, and percent error results for the head and trunk are presented in Tables 3 and 4, respectively. The IMU data strongly represented the criterion motion capture data for head ROM and peak rotational velocity across all conditions (ICC(A,1) > 0.9). RMSE across conditions remained low for head ROM but increased in the walking L/R and tandem walking L/R conditions for peak rotational head velocity. Despite the higher RMSE, the percent error for the head remained low across all conditions (<5%).

**Table 3.** Validity results comparing IMU to motion capture for head range of motion (ROM) and peak rotational velocity (ωp).


Intra-class Correlations Coefficients between the IMU and motion capture data for the trunk were stronger in the L/R direction (ROM ICC(A,1) > 0.9; peak rotational velocity ICC(A,1) > 0.9), than in the U/D direction (ROM ICC(A,1) = 0.580 to 0.907; peak rotational velocity ICC(A,1) = 0.436 to 0.787) across conditions. The reduced strength of the relationship is also mirrored in the RMSE and percent error scores for the U/D motions.

**Table 4.** Validity results comparing IMU to motion capture for trunk ROM and peak rotational velocity.


Bland-Altman plots did not indicate any clear patterns of bias. Examples are provided below for the head ROM and peak rotational velocity during walking with head turns (Figure 3).

**Figure 3.** Bland-Altman plots for walking conditions. Each participant is represented by a different color within the plot. (**A**) ROM, walking with head turns (L/R); (**B**) ROM, walking with head turns (U/D); (**C**) peak rotational velocity, walking with head turns (L/R); and (**D**) peak rotational velocity, walking with head turns (U/D). Solid lines represent the mean difference, and dashed lines represent ±1.96 × SD.
