Abstract
[Purpose] This study examined the eye movement characteristics of physical therapists during supervised level walking and stair negotiation and explored whether these characteristics differed according to clinical experience. [Participants and Methods] Twenty-seven physical therapists performed supervised level walking, and eighteen performed supervised stair negotiation while wearing a head-mounted eye tracker. Eye movement speed and binocular combined angle deviation were analyzed during each task. [Results] Right-eye movement speed during supervised level walking was descriptively higher in physical therapists with greater clinical experience, but statistically significant group differences were not confirmed after adjustment across eye movement outcomes. No statistically significant group differences in right-eye movement speed or binocular combined angle deviation were observed during supervised stair negotiation. [Conclusion] This exploratory study did not confirm clear experience-related differences in eye movement characteristics during supervised locomotion. The findings provide preliminary information on visual monitoring during supervised mobility tasks and highlight the need for further studies with larger samples and standardized task conditions.
Key words: Eye tracking, Physical therapists, Gait supervision
INTRODUCTION
Safe supervision during walking and stair negotiation requires physical therapists to continuously observe patient movements, anticipate instability, and respond appropriately to changes in balance and posture. In clinical practice, these observational skills are essential for preventing falls and providing effective gait supervision. Stair negotiation is especially demanding because it requires precise lower-limb control and is associated with a greater risk of instability than level walking1,2,3). Therefore, the way physical therapists visually monitor patients during supervised walking and stair negotiation may be an important component of their clinical skills. In physical therapy education, observational skills are often developed through repeated clinical experience and supervision. However, these skills are frequently taught and evaluated subjectively, based on instructors’ clinical impressions. Although experienced therapists may differ from less experienced therapists in how they detect movement cues, the objective characteristics of these differences remain unclear. Previous studies in sports and surgery have suggested that experts use visual information more efficiently than novices4, 5). In physical therapy, studies using eye tracking have begun to examine gaze behavior during movement observation and gait rehabilitation6,7,8). These studies indicate that gaze analysis may provide useful information for understanding observational skills in physical therapists.
Most previous studies have examined visual behavior during independent locomotion or during the observation of movement tasks, rather than during actual supervised mobility. Supervised walking and stair negotiation differ from simple observation because the therapist must monitor the patient while maintaining an appropriate position for safety. In this situation, visual monitoring is directly linked to clinical judgment and risk management. However, little is known about whether physical therapists with greater clinical experience exhibit different eye movement characteristics during supervised walking and stair negotiation.
Clarifying eye movement characteristics during supervised mobility may help identify objective features of observational skills in physical therapy. Such information may provide a basis for future research on visual monitoring and observational skill development during supervised walking and stair negotiation. Therefore, this study examined the eye movement characteristics of physical therapists during supervised level walking and stair negotiation and explored whether these characteristics differed according to clinical experience.
PARTICIPANTS AND METHODS
The participants were licensed physical therapists recruited from clinical and educational institutions. Initially, 30 physical therapists were scheduled for the supervised level walking task and 20 for the supervised stair negotiation task; however, three and two participants, respectively, did not participate because of cancellations on the day of the experiment, and no data were obtained from them. Therefore, twenty-seven physical therapists participated in the supervised level walking task. Their mean age was 30.2 ± 5.0 years, mean weight was 61.7 ± 11.1 kg, mean height was 168.1 ± 9.1 cm, and their mean clinical experience was 8.3 ± 4.9 years. For the supervised stair negotiation task, 18 physical therapists were included. Their mean age was 31.9 ± 5.1 years, mean weight was 60.3 ± 10.7 kg, mean height was 169.3 ± 7.2 cm, and their mean clinical experience was 10.0 ± 5.0 years. All participants had prior clinical experience in providing gait and stair supervision to patients. Physical therapists with visual impairments that could not be corrected with lenses or with neurological or musculoskeletal conditions affecting eye movements were excluded.
To explore experience-related differences in eye movement characteristics, participants were categorized into two groups based on years of clinical experience: less than 6 years and 6 years or more. This cut-off was operationally defined to distinguish physical therapists who had completed the typical early-career training phase from those with more established clinical practice. In this study, five years of clinical experience was considered a professional development milestone because postgraduate education programs for physical therapists in Japan are typically structured over a five-year period9), and the World Physiotherapy Education Framework identifies the initial years after licensure as a critical transition phase for competency development10). Therefore, physical therapists in their sixth year of clinical practice, who had completed at least five full years of clinical experience, were classified into the ≥6 years group, whereas those who had not yet completed five full years of clinical practice were classified into the <6 years group. These thresholds were used for exploratory purposes and were not intended to represent strict competency levels.
The supervised level walking experiment was approved by the Research Ethics Committee of Bunkyo Gakuin University (approval number: 2023-0003). The supervised stair negotiation experiment was approved by the Ethics Review Committee of Waseda University (approval number: 2024-256). Written informed consent was obtained from all participants.
Eye movements were recorded using a head-mounted eye-tracking system (Talk Eye Lite, Takei Scientific Instruments, Niigata, Japan). This device is a goggle-type eye-tracking system that uses a USB camera for eye detection and field-of-view recording and is connected to a processing computer. The system records binocular eye movements while allowing free head movements, making it suitable for natural supervised movement tasks. The eye-tracking data were sampled at 30 Hz. Before data collection, a standard calibration procedure was performed for each participant according to the manufacturer’s instructions.
The primary eye movement parameters were right-eye movement speed and binocular combined angle deviation. The binocular combined angle deviation represented the dispersion of gaze positions and was calculated separately for the horizontal and vertical directions. Smaller values indicate a narrower gaze distribution, whereas larger values indicate broader visual exploration. Eye movement speed was calculated as the average angular velocity of the right eye in degrees per second. However, eye dominance was not formally assessed. Previous studies have suggested an association between handedness and eye dominance11). Because most participants were presumed to be right-handed, right-eye data were used for the eye movement speed analysis.
The supervised level walking task was designed to simulate a common clinical situation in which a physical therapist visually observes a patient during level walking to anticipate instability and prevent falls. The physical therapists were not provided with any prior information about the patient and observed the gait for the first time during the trial.
A healthy adult male, aged 53 years, served as a simulated patient. To make the procedure as consistent as possible, he received pre-experiment training to use a cane with his left hand and to walk at a cadence of approximately one step per second. No specific disease diagnosis or clinical condition was assigned to the simulated patient. Actual walking speed during the measurement trials was not quantitatively recorded. The walking task was performed on a straight 10-m walkway, and one round trip was completed for each trial. Participants were instructed to provide stand-by supervision without physical contact. Because the simulated patient used a cane on the left side, all physical therapists naturally positioned themselves on the patient’s right side. They were asked to maintain a distance that allowed them to naturally monitor the simulated patient’s whole-body movement and prevent falls; thus, the exact distance was not constrained or measured.
No specific instructions regarding gaze behavior were given. Eye-tracking data were recorded continuously throughout the task. For the analysis, data were extracted beginning at the third foot contact of the simulated patient and continuing for 350 consecutive data points, corresponding to approximately 11.7 s at 30 Hz. The first and second steps were excluded because gait immediately after initiation may involve greater variability. The mean values of right-eye movement speed and binocular combined angle deviations in the horizontal and vertical directions were calculated over this interval.
The supervised stair negotiation task was designed to simulate a clinically challenging situation associated with an increased fall risk. The same simulated patient ascended and descended a staircase at a self-selected and comfortable pace while being supervised by the participant. The staircase consisted of 13 steps, with a width of 1.35 m, tread depth of 25.0 cm, and riser height of 17.0 cm. One complete round trip, including stair ascent and descent, was performed. Participants were instructed to provide stand-by supervision without physical contact and to maintain a position that allowed continuous visual monitoring of the simulated patient’s lower-limb movements and overall balance. No specific instructions regarding gaze behavior were provided. Stair ascent and descent were analyzed separately because these tasks have different biomechanical and balance-control demands1,2,3). For analysis, eye-tracking data were extracted from the period between initial contact on the first step to toe-off from the third step of the simulated patient for both ascent and descent. This segment was selected to ensure consistent measurement conditions across participants because camera stability could not be guaranteed over longer segments during stair negotiation. The mean values of right-eye movement speed and binocular combined angle deviation were calculated separately for stair ascent and descent.
The eye-tracking data were inspected before analysis to confirm data quality and completeness. Segments containing recording errors or signal losses that prevented reliable calculation of the predefined parameters were excluded. The normality of each variable was assessed using the Shapiro–Wilk test. Because at least one outcome variable violated the assumption of normality, nonparametric analyses were used. For supervised level walking, group differences between the <6 years and ≥6 years groups were examined using the Mann–Whitney U test. For supervised stair negotiation, the association between years of clinical experience and eye movement parameters was examined using Spearman’s rank correlation coefficients. Exploratory group comparisons were also performed using the Mann–Whitney U test with the same 6-year experience cutoff. For group comparisons, the Bonferroni correction was applied across the three eye movement outcomes within each task: right-eye movement speed, horizontal binocular combined angle deviation, and vertical binocular combined angle deviation. Effect size (r) was calculated as |Z|/√N. Data are presented as medians and interquartile ranges. Statistical analyses were performed using IBM SPSS Statistics (IBM Corp., Armonk, NY, USA). Statistical significance was set at p<0.05.
RESULTS
A total of 27 physical therapists were included in the analysis of the supervised level walking task. Twelve participants had less than 6 years of clinical experience, and 15 participants had 6 years or more of clinical experience. Right-eye movement speed was higher in the ≥6 years group than in the <6 years group. The median values were 162.5 deg/s [interquartile range, 92.1–214.9] in the ≥6 years group and 69.0 deg/s [37.9–116.0] in the <6 years group. This difference reached statistical significance before correction (U=42, p=0.020), but did not remain significant after Bonferroni correction across the three eye movement outcomes (adjusted p=0.061). The effect size was moderate (r=0.45). No significant group differences were observed in binocular combined angle deviation in either the horizontal or vertical direction. The median horizontal binocular combined angle deviation was 16.1° [11.1–25.7] in the <6 years group and 15.1° [13.2–19.5] in the ≥6 years group. The median vertical binocular combined angle deviation was 19.0° [12.4–22.5] in the <6 years group and 14.0° [10.8–17.3] in the ≥6 years group. The descriptive values for supervised level walking are shown in Table 1.
Table 1. Eye movement parameters during supervised level walking.
| Outcome | <6 years (n=12) Median [IQR] |
≥6 years (n=15) Median [IQR] |
| REMS (deg/s) | 69.0 [37.9–116.0] | 162.5 [92.1–214.9] |
| BCAD horizontal (deg) | 16.1 [11.1–25.7] | 15.1 [13.2–19.5] |
| BCAD vertical (deg) | 19.0 [12.4–22.5] | 14.0 [10.8–17.3] |
Values are presented as median [IQR: interquartile range]. REMS: right-eye movement speed; BCAD: binocular combined angle deviation.
A total of 18 physical therapists were included in the analysis of the supervised stair negotiation task. Five participants had less than 6 years of clinical experience, and 13 participants had 6 years or more of clinical experience. During stair ascent, right-eye movement speed was 80.0 deg/s [75.6–273.2] in the <6 years group and 77.0 deg/s [35.2–155.5] in the ≥6 years group. No significant group differences were observed (adjusted p=1.000; r=0.24). The horizontal binocular combined angle deviation was 22.1° [18.7–25.3] in the <6 years group and 18.4° [15.7–20.4] in the ≥6 years group. The vertical binocular combined angle deviation was 13.2° [13.1–14.0] in the <6 years group and 15.0° [11.6–16.1] in the ≥6 years group. No significant group differences were observed for these parameters after correction.
During stair descent, right-eye movement speed was 64.6 deg/s [58.0–350.6] in the <6 years group and 60.6 deg/s [34.5–92.3] in the ≥6 years group. The horizontal binocular combined angle deviation was 18.5° [16.5–19.9] in the <6 years group and 18.4° [8.3–24.8] in the ≥6 years group. The vertical binocular combined angle deviation was 14.2° [9.9–27.5] in the <6 years group and 13.0° [7.8–19.0] in the ≥6 years group. No significant group differences were observed during stair descent after Bonferroni correction across the three eye movement outcomes. The descriptive values for stair ascent and descent are presented in Table 2.
Table 2. Eye movement parameters during supervised stair negotiation.
| Stair phase | Outcome | <6 years (n=5) Median [IQR] |
≥6 years (n=13) Median [IQR] |
| Ascent | REMS (deg/s) | 80.0 [75.6–273.2] | 77.0 [35.2–155.5] |
| BCAD horizontal (deg) | 22.1 [18.7–25.3] | 18.4 [15.7–20.4] | |
| BCAD vertical (deg) | 13.2 [13.1–14.0] | 15.0 [11.6–16.1] | |
| Descent | REMS (deg/s) | 64.6 [58.0–350.6] | 60.6 [34.5–92.3] |
| BCAD horizontal (deg) | 18.5 [16.5–19.9] | 18.4 [8.3–24.8] | |
| BCAD vertical (deg) | 14.2 [9.9–27.5] | 13.0 [7.8–19.0] |
Values are presented as median [IQR: interquartile range]. REMS: right-eye movement speed; BCAD: binocular combined angle deviation.
Overall, statistically significant group differences in the selected eye movement measures were not confirmed after Bonferroni correction during either supervised level walking or supervised stair negotiation.
DISCUSSION
This study examined the eye movement characteristics of physical therapists during supervised level walking and stair negotiation and explored whether these characteristics differed according to clinical experience. Right-eye movement speed during supervised level walking was higher in the ≥6 years group before correction; however, this difference did not remain significant after Bonferroni correction across the three eye movement outcomes. No statistically significant group differences were observed during supervised stair ascent or descent. Therefore, the present exploratory analysis did not provide definitive evidence of experience-related differences in the selected eye movement measures.
During supervised level walking, right-eye movement speed was descriptively higher in physical therapists with greater clinical experience; however, this group difference did not remain statistically significant after Bonferroni correction. Binocular combined angle deviation did not differ significantly between the groups. Therefore, the level walking findings should be interpreted as preliminary and should not be considered definitive evidence of experience-related differences. Previous studies have suggested that experts differ from novices in terms of the efficiency and selectivity of visual information processing4). In physical therapy, eye-tracking studies have indicated that gaze behavior may provide useful information for understanding observational skills during movement analysis6,7,8). However, the present study differed from previous studies because therapists were required to move concurrently with the simulated patient and provide close supervision for fall prevention. These task characteristics may have influenced the selected eye movement measures, but this interpretation remains speculative because task difficulty and perceived fall risk were not quantitatively assessed.
The descriptive difference in right-eye movement speed during supervised level walking may indicate that this measure is worth further examination in future studies. During gait supervision, physical therapists must monitor multiple body regions and environmental information while maintaining patient safety. However, because the adjusted analysis did not confirm a statistically significant group difference, practical implications cannot be determined from the present results. During supervised stair negotiation, no statistically significant group differences were observed after correction. Stair ascent and descent require greater balance control and more precise foot placement than level walking1,2,3). In the present task, participants may have needed to monitor similar movement cues to maintain safety. However, because therapist-patient distance, perceived fall risk, and actual patient movement were not quantitatively assessed, this interpretation should be considered speculative.
Another possible explanation is that the eye movement parameters were not sufficient to capture all aspects of visual monitoring during stair negotiation. This study analyzed eye movement speed and binocular combined angle deviations. However, experience-related differences may appear in other dimensions, such as gaze timing relative to the patient’s movement, fixation location, predictive gaze behavior, or coordination between gaze behavior and therapist positioning. Therefore, the absence of significant differences during stair negotiation does not necessarily indicate that clinical experience does not influence visual monitoring. Rather, this suggests that more detailed analyses are required to evaluate visual behavior during complex supervised mobility tasks.
The present study has several limitations. First, the sample size was small, particularly for stair negotiation, and an a priori sample size calculation was not performed. Therefore, the study may have had limited statistical power to detect small or moderate group differences. Second, eye movement speed was calculated from right-eye data, and ocular dominance was not assessed. Although right-eye data were used consistently across participants and task conditions, individual differences in ocular dominance may have influenced eye movement speed. Future studies should assess ocular dominance and examine whether dominant- and non-dominant-eye data yield similar results. Third, biomechanical or kinematic data were not collected. Therefore, the relationship between eye movement characteristics and actual patient movement could not be examined. In addition, although the simulated patient received pre-experiment training to reproduce a similar walking pace, actual walking speed during the measurement trials was not quantitatively recorded. Therefore, trial-to-trial variation in walking pace may have influenced the eye movement measures. Fourth, years of clinical experience were used as an indirect index of expertise. We did not collect detailed data on educational background, postgraduate clinical training, or prior experience related to gait assessment or gait supervision. These unmeasured variables may influence observational skills independently of years of clinical experience. Therefore, years of clinical experience should be interpreted as a pragmatic but indirect index of expertise, and future studies should examine these participant characteristics in more detail.
Finally, this study used a simulated patient model, and the findings may not be generalizable to patients with neurological or musculoskeletal impairments.
Future studies should include larger samples and combine eye-tracking data with kinematic, kinetic, or clinical performance measures. In addition, analysis of areas of interest, fixation timing, and gaze behavior relative to patient movement may provide more detailed information about how physical therapists visually monitor patients during supervised mobility. Such approaches may help clarify how observational skills develop and how they can be trained in physical therapy education.
In conclusion, this study did not confirm statistically significant experience-related differences in eye movement characteristics during supervised level walking or stair negotiation after adjustment. These findings provide preliminary information on visual monitoring during supervised mobility tasks and suggest that larger studies with more detailed task standardization and gaze analyses are needed.
Conference presentation
Parts of this study were presented at the World Physiotherapy Asia Western Pacific Regional Congress 2024 and the World Physiotherapy Congress 2025.
Funding
This work was supported by JSPS KAKENHI under Grant 23K11951.
Conflict of interest
The authors declare no conflicts of interest.
Funding Statement
This work was supported by JSPS KAKENHI under Grant 23K11951.
REFERENCES
- 1.Riener R, Rabuffetti M, Frigo C: Stair ascent and descent at different inclinations. Gait Posture, 2002, 15: 32–44. [DOI] [PubMed] [Google Scholar]
- 2.Startzell JK, Owens DA, Mulfinger LM, et al. : Stair negotiation in older people: a review. J Am Geriatr Soc, 2000, 48: 567–580. [DOI] [PubMed] [Google Scholar]
- 3.Zietz D, Johannsen L, Hollands M: Stepping characteristics and Centre of Mass control during stair descent: effects of age, fall risk and visual factors. Gait Posture, 2011, 34: 279–284. [DOI] [PubMed] [Google Scholar]
- 4.Mann DT, Williams AM, Ward P, et al. : Perceptual-cognitive expertise in sport: a meta-analysis. J Sport Exerc Psychol, 2007, 29: 457–478. [DOI] [PubMed] [Google Scholar]
- 5.Evans-Harvey K, Erridge S, Karamchandani U, et al. : Comparison of surgeon gaze behaviour against objective skill assessment in laparoscopic cholecystectomy—a prospective cohort study. Int J Surg, 2020, 82: 149–155. [DOI] [PubMed] [Google Scholar]
- 6.McDuff K, Benaim A, Wong M, et al. : Analyzing the eye gaze behaviour of students and experienced physiotherapists during observational movement analysis. Physiother Can, 2021, 73: 129–135. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Liu Z, Makino K, Nakamura M, et al. : Analysis of eye tracking of physiotherapist during walk rehabilitation. In: IECON 2019—45th Annual Conference of the IEEE Industrial Electronics Society, 2019, pp 6832–6837. [Google Scholar]
- 8.Suzuki R, Kurita Y: Characteristics of gaze tracking during movement analysis by therapists. J Phys Ther Sci, 2022, 34: 36–39. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Japanese Physical Therapy Association: Renewal of registered physical therapist. https://www.japanpt.or.jp/pt/lifelonglearning/new/registered/ (Accessed Mar. 25, 2026)
- 10.World Physiotherapy: Physiotherapist education framework. London: World Physiotherapy, 2021. [Google Scholar]
- 11.Bourassa DC, McManus IC, Bryden MP: Handedness and eye-dominance: a meta-analysis of their relationship. Laterality, 1996, 1: 5–34. [DOI] [PubMed] [Google Scholar]