Vestibular dysfunction, regardless of etiology, may result in a variety of signs such as gaze and postural instability. In addition, individuals with vestibular dysfunction may report reduced balance confidence, oscillopsia, and/or vertigo.1 The provision of efficacious treatments is critically important to reduce dizziness handicap, reduce motion sensitivity,2 improve gaze and postural stability, and improve dynamic visual acuity.3
A recent clinical practice guideline on the treatment of vestibular hypofunction not only summarized the evidence supporting treatment, but also clearly articulated specific gaps in our knowledge regarding treatment.4 While there is moderate to strong evidence supporting the efficacy of gaze and postural stability training to improve domains of disability in individuals with vestibular hypofunction, the current evidence that specifies dosage of gaze stability exercises is based on the protocols taken from a small number of human studies.3–5 In general, clinical recommendations are to perform gaze stability exercises with active head movements multiple times per day without additional clear guidance regarding systematic progression. As an example of the lack of clarity regarding progression, in their clinical practice guideline, Hall et al4 made the recommendation that, “Researchers should examine the impact of frequency, intensity, time, and type of exercises on rehabilitation outcomes. Researchers should determine the difficulty of exercises and how to progress patients in a systematic manner.”
Although documentation of treatment dosage is a valuable objective, there are few examples that have described the effective use of commercially available tools to document or guide treatment. In one example, Huang et al6 used the gyroscope integrated into an iPod (Apple Inc, Cupertino, California) to measure head movements and examine the adherence of individuals with vestibular hypofunction with their prescribed home gaze stabilization exercise program. While the iPod represents one option, the relatively recent clinical availability of wearable inertial measurement units (IMUs) provides an alternate objective technological tool to examine head movements in the clinic as well as in home and community settings.7 These wearable sensors can quantify the head motions associated with gaze stability exercises in a minimally invasive manner,5,6 allow for comparison of different movement tasks, and may document treatment progression over time.
In order for a wearable technology to be clinically useful for vestibular rehabilitation, it must have utility—that is it must be able to distinguish between groups exposed to varied head movement demands. In addition, it should be able to document head movement dosage (ie, head movement frequency, velocity, and amplitude) performed as part of gaze stability treatment. To examine the utility of IMUs in vestibular rehabilitation, we measured head kinematics during in-clinic treatments using a head-mounted IMU. Our first aim was to investigate the ability of using head-worn IMUs to differentiate between varied group exercise protocols. Specifically, we hypothesized that an IMU worn on the head would distinguish head movement behaviors between a group performing guided gaze and postural stabilization exercises (GPS) from a strength and aerobic exercise control group (SAE) where specific head movement exercises were not included. The second aim was to determine whether the IMUs would quantify the progression of gaze stability exercises within the GPS group over the course of a 6-week treatment. We hypothesized that the IMU worn on the head during an early, middle, and late treatment sessions by individuals in the GPS group would allow quantification of the progression of the frequency, velocity, and amplitude of head movements performed during yaw and pitch plane gaze stabilization exercises.
The participants in this study were participating in a parent trial of individuals with a diagnosis of multiple sclerosis (MS) and complaints of dizziness or a history of falls in the preceding year. Recruitment in the parent trial began in June 2018 from an accessible population of people with MS within the Salt Lake City metropolitan area. The instrumentation and algorithms used for data processing for the current study became available for regular use in December, 2018. All participants enrolled in the parent trial subsequent to that time were included. Participants were eligible if they were aged 18 to 75 years, had neurologist-diagnosed MS, had complaints of dizziness (Dizziness Handicap Inventory [DHI] >0), an Activities-Specific Balance Confidence Scale score less than 80%, or greater than 1 fall in the last year. They were also excluded if they had any other central or peripheral nervous system disorders. Lastly, because the parent study investigated in the influence of gaze and postural stability on individuals with only central nervous system (CNS)–mediated vestibular dysfunction, individuals with peripheral vestibular pathology were excluded. The absence of peripheral nervous system–mediated vestibular dysfunction and the presence of CNS-mediated vestibular dysfunction were confirmed by systematically gathering a focused history asking about signs and symptoms as well as treatment for prior peripheral vestibular pathology.8 This was followed by clinical testing to rule out peripheral causes of vestibular symptoms (benign paroxysmal positional vertigo and hypofunction tests). Next, we confirmed the presence of CNS-mediated oculomotor deficits (saccadic smooth pursuit, dysmetric saccades, and abnormal vestibular ocular reflex [VOR] or VOR cancellation) and/or a perverted head shake nystagmus (fixation removed).6 Lastly, we documented that deficits in the VOR gain and saccadic corrections measured via head impulse testing did not follow a pattern consistent with peripheral vestibular pathology.9 This testing was performed by 2 of the authors (A.F. and L.D.) as part of measures of disease severity and oculomotor behaviors. Although the participants were recruited from the parent trial, estimation of the sample size needed to limit type II error risk for these analyses was performed using data from Paul et al.10 Using the effect size calculated from the between-group comparisons of head velocity data, a preset α of 0.05, and a power of 0.80, the sample size was estimated to be 20 participants. Approval for this study was granted by the University of Utah Institutional Review Board (IRB_00104298).
Twenty-eight participants previously randomized to either the GPS group or the SAE group were used in this analysis. Although our intent was to include all participants from the parent trial, equipment availability and malfunctions only allowed us to gather data from 28 participants. Prior to participation, all participants signed an informed consent form and the overall consent for the parent trial included the use of wearable sensors during treatment. In both groups, participants were taken through 18 sessions of guided exercise over 6 weeks (3 times/week) at one clinical site. All treatments were provided in an individualized manner under the supervision of a trained research assistant. The research assistant also closely documented the timing, duration, frequency, and intensity of each exercise in a treatment log that allowed for examination of treatment fidelity and postprocessing of individual exercises.
The gaze stability exercises and the progression of difficulty of concurrent gaze and postural stability tasks have been used in previous research on aVOR gain recovery with vestibular rehabilitation.1,11 Briefly, gaze stability exercises included angular vestibular ocular reflex (aVOR) exercises with a stationary target (VORx1) in both yaw and pitch planes. VORx1 exercises involved participants standing 3 ft from an eye-level target and turning their head ∼30° (∼15° to each side or up and down of neutral rotation) while keeping their eyes fixed on the stationary target. Using a metronome on a smartphone, participants started their gaze stabilization exercises by turning their head in time with the metronome set at 60 beats per minute (bpm) (1 Hz) and the cadence was adjusted up or down depending on tolerance, with the goal of 120 bpm (2 Hz) or greater. To progress the velocity of head turns, participants were instructed to perform head turns as rapidly as possible while still maintaining visual focus on the target and tolerable levels of dizziness. In addition to attempting to progress the velocity and frequency of head movements during VORx1 exercises, the demands of maintaining postural stability during the VORx1 training was concurrently progressed (ie, normal stance → semi-tandem → full tandem → compliant surface). Participants in the GPS group received ∼15 minutes of active horizontal and vertical head motion gaze stability training per treatment session. The goal was to have participants perform repetitive 1-minute bouts of gaze stabilization with breaks between provided to allow any provoked symptom to subside. Participants in the SAE group were not prescribed any specific gaze stability exercises, and both groups performed strength training for the lower extremities using an commercially available leg press machine (Precor Inc, Woodinville, Washington) and aerobic endurance training using a NuStep upper and lower extremity dynamometer (NuStep Inc, Ann Arbor, Michigan). Details regarding the specific treatment progression of both types of exercise are provided in Loyd et al.12
Instrumentation for Treatment Monitoring
Using wearable IMUs (Opal Monitors, APDM Inc) with onboard accelerometers, gyroscopes, and magnetometers to track head movement characteristics, we collected data from a sensor worn over the forehead during the initial 15 minutes of an individual treatment visit at 3 timepoints in the study, early (week 1), middle (week 3), and late (week 6) while participants completed assigned group exercises. The initial 15 minutes was selected for analysis to capture the gaze stabilization exercise component of the GPS group, which always occurred first, while the SAE group would be completing lower extremity strengthening during that same period. Once gathered, sensor data were processed using a custom Matlab algorithm. This processing involved registration of the sensor to a global reference frame to ensure planar motion was consistent regardless of the angle of the sensor. Additionally, the signal from the sensor was filtered using a 6-Hz low-pass Butterworth filter. The individual responsible for processing the data was not blinded to group assignment or measurement period.
Head turns (yaw plane) and head nods (pitch plane) were identified based on peaks in angular velocity data in the yaw and pitch planes. A peak was considered a turn if it had (1) a velocity greater than 20°/s and (2) an amplitude greater than 35% of the mean amplitude of all peaks for that subject. Frequency of head turns was calculated as the number of head turns during the recorded period and dividing it by the total time. Frequency was used as a metric of the exposure per unit time that the participant had to head movements during the treatment. The peak velocity of head turns was calculated by identifying the highest velocity of each head turn regardless of direction. The amplitude of each head turn was identified by multiplying the angular velocity peak value by the width of the peak resulting in the angular rotation for each head turn. This resulted in an amplitude calculation that represented turns from the furthest in one direction to the furthest value in the other direction (ie, from the most rightward head position to the most leftward head position). The absolute values of each yaw and pitch head turn velocity and amplitude were calculated, and then overall means were calculated to represent the average peak velocity and amplitude of head turns occurring for each subject. For study aim 1 (examination of between-group differences), data from the 15 minutes of gaze stability exercises in the GPS group were compared with the comparable time in the SAE group. Data from the early, middle, and late periods were pooled and the means, standard deviations, and confidence intervals for each dependent variable were generated for each group.
For study aim 2 (the examination of GPS group progression), we used the same Matlab code as for aim 1 to isolate periods of active gaze stability training, namely VORx1 in the yaw and pitch planes. Using treatment log notes from the training session and visual identification, the periods of VORx1 in yaw and pitch plans were isolated separately and the frequency, average peak velocity, and average peak amplitude were calculated for each. These values were generated within the GPS group at each time period (early, middle, and late).
Prior to conducting statistical analysis, we calculated baseline descriptive statistics including mean and SD for age, sex (percent female), body mass index, DHI, and Expanded Disability Severity Scale (EDSS) scores and the dependent variables for both groups. Baseline differences in demographics between the groups for continuous variables were compared using independent-samples t tests, and the sex distributions were compared using the χ2. For study aim 1, general linear models for frequency, velocity, and amplitude, in both the yaw and pitch plane, were fit on the independent variable of the group. Significance for between-group differences was set at an α level of P < 0.05, and the t values, mean differences between the groups, and 95% confidence intervals were presented. For study aim 2, examining the within-group progression of VORx1 exercise, general linear regression models for frequency, velocity, and amplitude, in both the yaw and pitch planes, were fit on the time effect. When the omnibus test of the time effect reached statistical significance (P < 0.05) post hoc pairwise t tests with a Bonferroni adjustment were performed to test for specific differences in the timepoints.
Demographic and baseline functional status of the participants in both groups are included in Table 1.
Table 1 -
Descriptors Mean ± SD and Percentage of Female Participants for Both Groupsa
||GPS (n = 15)
||SAE (n = 13)
||56.1 ± 10.2
||50.5 ± 15.4
||28.3 ± 8.3
||24.7 ± 9.0
||3.9 ± 0.9
||3.5 ± 1.2
||38.5 ± 15.5
||42.2 ± 23.3
Abbreviations: BMI, body mass index; DHI, Dizziness Handicap Inventory; EDSS, Expanded Disability Severity Scale; GPS, gaze and postural stability; SAE, strength and aerobic exercise.
aEDSS and DHI scores were recorded at the baseline assessment, prior to beginning training for all participants.
bSignificant between-group differences (P < 0.05).
There was a significant difference between the groups for the velocity and amplitude of yaw (P < 0.001 and P < 0.001, respectively) and pitch (P < 0.001 and P < 0.001, respectively) head turns, with the GPS group demonstrating greater mean velocity and amplitude of head turns. In contrast, there was no significant between-group difference in the frequency of head turns in the yaw (P = 0.2) or pitch planes (P = 0.3) (Table 2).
Table 2 -
Mean ± SD for the GPS and SOC Groups and the Mean Difference Between the 2 Groups
||Mean Difference (95% CI)
|Frequency, head turn/s
||0.76 ± 0.2
||0.69 ± 0.2
||−0.07 (−0.03 to 0.2)
||24.2 ± 8.0
||15.1 ± 5.7
||−9.1 (−12.0 to −5.8)a
||87.8 ± 24.3
||62.7 ± 25.1
||−25.1 (−36.1 to −14.1)a
|Frequency, head turn/s
||0.72 ± 0.2
||0.78 ± 0.3
||0.04 (−0.2 to 0.1)
||15.0 ± 5.3
||9.3 ± 3.7
||−5.7 (−7.8 to −3.7)a
||61.7 ± 18.0
||41.7 ± 14.4
||−20.0 (−27.3 to −12.7)a
Abbreviations: CI, confidence interval; GPS, gaze and postural stability; SAE, strength and aerobic exercise.
aSignificant mean difference between the 2 groups (P < 0.05).
When examining the periods of VORx1 exercises in the GPS group, a significant time effect was observed for the frequency of head turns in both the yaw (t value = 5.9; P < 0.001) and pitch (t value = 5.9; P < 0.001) planes. Post hoc testing revealed that in the yaw plane there was a significant change from the early to middle (P = 0.004), middle to late (P < 0.001), and early to late (P < 0.001) (Figure 1A). In the pitch plane there was a significant time effect from early to middle (P = 0.03), middle to late (P = 0.005), and early to late (P < 0.001) (Figure 2A). The time effect was not significant for the measures of velocity in yaw (t value = 0.5; P = 0.6) or pitch (t value = 0.7; P = 0.5) (Figures 1 and 2B). Similarly, the time effect was not found to be significant for amplitude in the yaw plane (t value = −1.5; P = 0.11) or the pitch plane (t value = −0.7; P = 0.4).
While clinical practice guidelines suggest that the dosage and type of gaze stabilization exercises should be tested to determine their effect on clinical outcomes,4 a precursor to such studies is the objective measurement of head kinematics during prescribed exercises.5 Wearable sensors provide an opportunity to study the dosage (ie, frequency, velocity, and amplitude) of head movement to establish goals for the delivery of effective gaze stability exercises. In this study, we sought to characterize the utility of IMUs to measure head kinematics during gaze stabilization exercises. As hypothesized in aim 1, the IMUs captured significant differences in head turn velocity and amplitude between a group of people performing gaze stability exercises and a group performing a strength and endurance exercise program that did not prioritize head movements. In addition, the IMUs worn during treatment allowed for the detection of significant increases in the frequency of head movements over a 6-week training period within the GPS group, while also demonstrating nonsignificant increases in velocity and decreases in amplitude. Such findings support aim 2 by documenting the utility of IMUs as a means to characterize the dosage of gaze stabilization exercises. Taken together, these results may provide a foundation for using IMUs as a means to objectively measure patients' dosage and the quality of the performance of gaze stabilization exercises.
Confirming the Dosage of Gaze Stabilization Exercises in 2 Ways
The 2 treatment groups in the parent study were designed to differ primarily in the dosage of vestibular stimulating exercises prescribed. While the exercises for each group were explicitly planned, the IMUs quantified specific kinematic features to confirm the intent of the exercises. Based on previous animal and human research regarding gaze stabilization via VOR mechanisms, we characterized the dosage of head movement kinematics via the frequency, velocity, and amplitude of head movements.4,13 Within the context of this study, it was expected that the GPS exercises would provide a greater physiologic stimulus than the activities of the SAE group, which did not prioritize head movements, through greater exposure to gaze instability error signals for the cerebellum and other CNS structures to recognize and attempt to correct.14
For the between-group comparison, we were primarily interested in the dosage components of head movement velocity and amplitude. When comparing groups, we measured head kinematics of each group during equivalent length periods as part of their treatment sessions. The main difference between the exercise programs during the measured period was that the GPS group participants performed VORx1 exercises. The accumulation of hundreds of repetitions of gaze stabilization exercises at a goal velocity substantially higher and a movement amplitude larger than habitual head movements accounted for the observed increases in yaw and pitch plane movement velocity and amplitude. In contrast, since head movement frequency was calculated as the number of head turns per unit time and the gaze stabilization exercises of the GPS group comprised approximately 20% of the overall measurement period (ie, ∼15 minutes of gaze stability, ∼15 minutes of postural stability, and ∼45 minutes of strength and endurance), it was not surprising that head movement frequency was not significantly different between groups.
For the within-group comparisons, our exploration of the utility of the IMUs was focused on movement frequency. Given that our frequency outcome measure was calculated as the number of head movements per unit time, the IMU measurement of VORx1 exercises over repeated 2-minute bouts allowed for a more granular examination of the change in frequency during these focused periods of time. The target threshold frequency of head movement during treatment was at or greater than 2 Hz (120 bpm on the metronome) due to the common finding that the head movement frequency during human gait and other activities is close to 2 Hz.15,16 The observed increases in head movement frequency over the 6 weeks of treatment to frequencies that approximate functional mobility support the utility of the IMUs for quantifying this aspect of gaze stabilization exercise dosage.
The measurement of velocity and amplitude of the GPS group over the treatment period provided the final means of establishing the utility to IMUs for treatment monitoring. The goal for angular head velocity during gaze stabilization exercises was 120°/s to emphasize VOR gaze stabilization and to generate gaze position errors that would drive saccadic corrections.17–20 We found the mean peak head turning velocities in the yaw and pitch planes during VORx1 exercises in the early session were 100.2°/s and 79.3°/s, respectively, and progressed to 113.6°/s and 79.5°/s, respectively, over the course of 6 weeks (Table 3). Although not a significant change over time, the head movement velocities approximated those velocities shown to be sufficient to force the nervous system to stabilize gaze in various vestibular physiology studies.17–20 Although the mean values observed by Huang et al6 (yaw = 140°/s ± 35°/s; pitch = 101°/s ± 29°/s) were higher in their study of individuals with peripheral vestibular hypofunction, the presence of multisensory, motor, and coordination deficits associated with MS in the current cohort may have limited the progression of velocity during the gaze stability exercises. In addition, these increases in velocity occurred concurrently with the progression of postural demands during the VOR exercises.
Table 3 -
Mean ± SD for Yaw and Pitch Plane VOR Periods in GPS Group
||1.6 ± 0.4
||1.6 ± 0.5
||2.3 ± 0.5
||2.2 ± 0.5
||3.1 ± 0.6
||3.0 ± 0.7
||44.4 ± 34.1
||28.0 ± 11.8
||31.8 ± 12.9
||26.3 ± 11.4
||26.0 ± 9.2
||21.7 ± 13.6
||100.2 ± 35.3
||79.3 ± 28.8
||112.7 ± 34.9
||88.8 ± 27.3
||113.6 ± 35.7
||79.5 ± 22.7
Abbreviations: GPS, gaze and postural stability; VOR, vestibular ocular reflex.
Lastly, amplitude-based metrics provided a picture of how participants perform their exercises. Although mean values are important, knowing the range of amplitude values provides clearer insight about the quality of the exercises. For example, Huang et al6 report 124° as the high range of yaw plane head amplitude during prescribed gaze stabilization exercises in one subject.6 Such a movement amplitude makes it anatomically impossible for that individual to maintain their gaze stabilized on a target. In our sample, participants were found to have a mean yaw plane amplitude of head turns of 44.4° at the early timepoint during VORx1 exercises, but demonstrated the expected reduction of head turn amplitude (31.8° at middle and 26.0° at late) necessary to accommodate for increasing frequency and velocity of head turns.
A Step Toward Clinical Use
The data presented here are a step toward the clinical use of IMUs to quantify head kinematics during in-clinic gaze stabilization exercises. Clinicians can rely on fairly “low tech” devices such as a metronome to help drive increases in the frequency of head turns, but lack tools to accurately capture the velocity and amplitude of these head turns. Regular use of IMUs would allow rehabilitation practitioners a tool to complement subjective assessments of the intensity achieved during training and objectively document the progression of treatment dosage (frequency, velocity, and amplitude). Future work is needed to develop clinically useable interfaces that can track these outcomes in real time, easing the implementation of IMUs into practice, and ultimately informing clinicians and patients on the quality and progress of treatment.
Limitations and Future Directions
Although we examined head kinematics in detail, this study did not measure eye movements during the gaze stabilization exercises. Future research should concurrently measure head and eye motion during prescribed exercises. Additional limitations to this work include a relatively small sample size of participants in each group and varied measured treatment times. This sample also resulted in significantly different proportions of female participants in each group. Despite these differences, sex was not fit as a covariate when comparing the 2 groups, as previous studies have not demonstrated significant sex differences in the VOR.21 Furthermore, in this study the head kinematics were measured in a sample of people with MS with varied motor and sensory deficits that increased the variability of our measures. A more tightly controlled examination of the performance of gaze stability exercise with the use of IMUs in people with isolated unilateral vestibular loss would add additional support to our findings.
While exercises performed in the clinic are one component of effective treatments, successful vestibular rehabilitation is predicated on the assumption that patients adhere to a home gaze stabilization exercise program. Previous research has revealed a significant lack of adherence with prescribed exercise dosages.5 Future research that uses IMUs to measure adherence, as well as monitor head movement progression during community activity, would be a substantial advancement in vestibular rehabilitation treatment.3,10,22,23
Wearable IMUs were used to examine the kinematics of head movements during rehabilitation treatments. As hypothesized, the IMUs captured significant between-group differences, as well as detecting the progression of head movements over a 6-week training. Such findings may provide a support for the utility of IMUs as a means to measure persons' head kinematics during gaze stabilization exercises and provide an objective means of documenting and progressing the dosage of gaze stabilization exercises during vestibular rehabilitation.
1. Han BI, Song HS, Kim JS. Vestibular rehabilitation
therapy: review of indications, mechanisms, and key exercises. J Clin Neurol. 2011;7(4):184–196. doi:10.3988/jcn.2011.7.4.184.
2. Gaikwad SB, Johnson EG, Nelson TC, et al. Effect of gaze stability
exercises on chronic motion sensitivity: a randomized controlled trial. J Neurol Phys Ther. 2018;42(2):72–79.
3. Lacour M, Bernard-Demanze L. Interaction between vestibular compensation mechanisms and vestibular rehabilitation
therapy: 10 recommendations for optimal functional recovery. Front Neurol. 2015;5:285.
4. Hall CD, Herdman SJ, Whitney SL, et al. Vestibular rehabilitation
for peripheral vestibular hypofunction: an evidence-based clinical practice guideline: from the American Physical Therapy Association Neurology Section. J Neurol Phys Ther. 2016;40(2):124–155.
5. Huang K, Sparto PJ, Kiesler S, Smailagic A, Mankoff J, Siewiorek D. A technology probe of wearable in-home computer-assisted physical therapy. In: Association for Computing Machinery (ACM); 2014:2541–2550. doi:10.1145/2556288.2557416.
6. Bonato P. Advances in wearable technology
and applications in physical medicine and rehabilitation. J Neuroeng Rehabil. 2005;2(1):2. doi:10.1186/1743-0003-2-2.
7. Rine RM, Schubert MC, Whitney SL, et al. Vestibular function assessment using the NIH Toolbox. Neurology. 2013;80(11, suppl 3):S25–S31.
8. Zuma E, Maia FC, Cal R, D'Albora R, Carmona S, Schubert MC. Head-shaking tilt suppression: a clinical test to discern central from peripheral causes of vertigo. J Neurol. 2017;264(6):1264–1270.
9. Fetter M, Dichgans J. Vestibular neuritis spares the inferior division of the vestibular nerve. Brain. 1996;119(3):755–763.
10. Paul SS, Dibble LE, Walther RG, Shelton C, Gurgel RK, Lester ME. Characterization of head-trunk coordination deficits after unilateral vestibular hypofunction using wearable sensors. JAMA Otolaryngol Neck Surg. 2017;143(10):1008–1014.
11. Schubert MC, Migliaccio AA, Clendaniel RA, Allak A, Carey JP. Mechanism of dynamic visual acuity recovery with vestibular rehabilitation
. Arch Phys Med Rehabil. 2008;89(3):500–507.
12. Loyd BJ, Fangman A, Peterson DS, et al. Rehabilitation to improve gaze and postural stability in people with multiple sclerosis
: study protocol for a prospective randomized clinical trial. BMC Neurol. 2019;19(1):119.
13. Clendaniel RA, Lasker DM, Minor LB. Differential adaptation of the linear and nonlinear components of the horizontal vestibuloocular reflex in squirrel monkeys. J Neurophysiol. 2002;88(6):3534–3540.
14. Shelhamer M, Tiliket C, Roberts D, Kramer PD, Zee DS. Short-term vestibulo-ocular reflex adaptation in humans. Exp brain Res. 1994;100(2):328–336.
15. Murray MP, Drought AB, Kory RC. Walk patterns of normal men. J Bone Joint Surg A. 1964;46:335–360.
16. MacDougall HG, Moore ST. Marching to the beat of the same drummer: the spontaneous tempo of human locomotion. J Appl Physiol. 2005;99(3):1164–1173.
17. Demer JL, Crane BT. Vision and vestibular adaptation. Otolaryngol Neck Surg. 1998;119(1):78–88.
18. Crane BT, Demer JL. Human gaze stabilization during natural activities: translation, rotation, magnification, and target distance effects. J Neurophysiol. 1997;78(4):2129–2144.
19. Pulaski PD, Zee DS, Robinson DA. The behavior of the vestibulo-ocular reflex at high velocities of head rotation. Brain Res. 1981;222(1):159–165.
20. Hirvonen TP, Pyykkö I, Aalto H, Juhola M. Vestibulo-ocular reflex function as measured with the head autorotation test. Acta Otolaryngol. 1997;117(5):657–662.
21. Smith PF, Agrawal Y, Darlington CL. Sexual dimorphism in vestibular function and dysfunction. J Neurophysiol. 2019;121(6):2379–2391.
22. Paul SS, Dibble LE, Walther RG, Shelton C, Gurgel RK, Lester ME. Reduced purposeful head movements during community ambulation following unilateral vestibular loss. Neurorehabil Neural Repair. 2018;32(4/5):309–316.
23. Lacour M, Helmchen C, Vidal PP. Vestibular compensation: the neuro-otologist's best friend. J Neurol. 2016;263(1):54–64.