Study at a glance
Summary
Disclaimer: The following summary details independent academic research. Theia3D is intended for research, sports performance and educational use only. It is not cleared as a medical device and is not intended to be used for the diagnosis, treatment, mitigation or prevention of any disease, injury or medical condition.
Editor's note: Theia3D version v2021.2.0.1675, as stated in the paper.
Study Overview
Design
A test-retest reliability study assessing the day-to-day reproducibility of Theia3D-derived joint angles during level walking, ramp ascent, ramp descent, stair ascent, and stair descent, with each participant completing identical testing sessions on two separate days at least 24 hours apart.
Participants
Twenty-one healthy adults (14 men, 7 women; mean age 31.1 years; mean height 1.67 m; mean weight 63.3 kg) recruited from staff at the National Center for Geriatrics and Gerontology, excluded if they had a neuromuscular or musculoskeletal condition that could interfere with walking performance.
Data Collection
Testing took place in a purpose-built "living laboratory" designed to simulate real home and outdoor environments, including a bedroom, living room, kitchen, bathroom, entrance, outdoor terrain, a 3.25-meter ramp at a 7-degree gradient, and two stair configurations. On each of two testing days, participants completed five repetitions each of a 7-meter level walk, ramp ascent, ramp descent, and ascent and descent of a four-step, 15-centimeter-height staircase, starting on their right foot at their usual walking pace without using the available handrails. Twenty-seven synchronized Sony RX0M2 cameras recorded each session at 60 Hz and were processed through Theia3D to generate hip, knee, and ankle joint angles in all three planes of motion, time-normalized to the gait cycle.
Analysis
Day-to-day reliability was assessed using two complementary methods: full-curve root-mean-square difference (RMSD) analysis across each joint angle waveform, and discrete-event analysis using the standard error of measurement (SEM) at specific, representative points in the gait cycle for level walking, ramp ascent, and ramp descent (stair ascent and descent were excluded from this discrete-event analysis because toe-off could not be reliably determined for these tasks). Measurement errors between 2 and 5 degrees were considered acceptable based on established reliability criteria from the gait analysis literature.
Key Findings
- Day-to-day measurement error stayed below the 5-degree acceptability threshold for the large majority of joint angles across all five tasks tested. SEM values for level walking, ramp ascent, and ramp descent ranged from 0.71 to 4.37 degrees across all joints and planes, and full-curve RMSD values across all five tasks ranged from 1.2 to 5.7 degrees.
- Only two specific measures exceeded the 5-degree threshold, and only modestly: knee flexion-extension RMSD during ramp ascent (5.07 degrees) and during stair ascent (5.64 degrees). Otherwise, all other joint angles and planes across every task tested fell within the acceptable range.
- This is the first published study to report reliability data for stair ascent and descent specifically using markerless motion capture. Stair gait showed slightly higher RMSD values (1.8 to 5.7 degrees for ascent; 1.5 to 4.5 degrees for descent) than level walking (1.2 to 2.4 degrees) or ramp walking, which the authors attribute to the greater complexity and movement variability inherent to stair negotiation.
- Ramp and stair tasks produced substantially larger joint range-of-motion changes compared with level walking, reflecting distinct postural strategies for each task. Stair ascent increased hip range of motion by roughly 10.6 degrees and knee range of motion by roughly 17.6 degrees compared with level walking, while stair descent decreased hip range of motion by roughly 18.7 degrees and increased ankle range of motion by roughly 14.5 degrees compared with level walking.
- Measurement reliability for level walking in this living-laboratory setting was consistent with reliability figures previously reported for Theia3D in more conventional laboratory environments, despite the living laboratory's more complex, obstacle-containing environment, including handrails, walls, and a divided indoor/outdoor space, that the authors note could plausibly have challenged camera-based tracking.
What This Means for Movement Science Researchers and Assistive Technology Teams
For researchers and developers working on assistive technology, robotics, or movement research aimed at real-world mobility, this study offers a clear, practically useful answer: Theia3D's day-to-day measurement reliability, already established for level walking, extends to ramp and stair ascent and descent as well, with the vast majority of joint angle measures falling within an acceptable reliability range even in a complex, home-like testing environment. The only joint angle that consistently showed borderline reliability was knee flexion-extension during ascent tasks specifically, worth keeping in mind for any application that depends heavily on precise knee kinematics during stair or ramp climbing. More broadly, this kind of simulated living-space testing environment offers a practical way to study how people's gait actually adapts to the varied terrain of everyday life, supporting research into fall-prevention strategies, assistive device design, and robotics applications intended for real-world environments rather than a controlled, flat-ground laboratory alone.
The Role of Theia3D in Movement Analysis
Theia3D was the markerless motion capture system evaluated throughout this study, processing video from a 27-camera array to generate joint kinematics across five distinct locomotion tasks within a simulated home and outdoor environment. By establishing Theia3D's day-to-day reliability not just for level walking but for ramp and stair ascent and descent specifically, including the first published reliability data for stair gait using markerless motion capture, this study extends the evidence base supporting Theia3D's use for research and technology development focused on the full range of everyday movement, not level walking alone.
Study Limitations and Future Directions
The authors note that clothing was not standardized across the two testing sessions, and while thirteen of the twenty-one participants wore the same clothing on both days, this was not directly verified as a factor, so clothing consistency may have contributed to the reliability results obtained. The algorithm used to detect foot contact and foot lift events relies on the positional relationship between the feet, which may not generalize well to more irregular gait patterns, a consideration for extending this work to populations whose gait differs meaningfully from a typical pattern. This study's sample was limited to healthy young adults, and the authors note that no verification was conducted in older adults or individuals with gait-related conditions, the populations most likely to benefit from the kind of real-world mobility assessment this living laboratory is designed to support, so further testing in these groups is an important next step. The study also did not include a concurrent marker-based comparison, since its focus was specifically on between-day reliability rather than systematic or random measurement error relative to a marker-based reference; the authors note that a simultaneous marker-based comparison would provide additional, complementary validation. Finally, the authors note that while reliability was established using RMSD and SEM, determining whether an observed change in gait parameters is practically meaningful also requires establishing a minimal practically important difference, which has not yet been established for the specific tasks and outcomes studied here, and they identify this as an important direction for future research.
Citation: Shimotori, D., Kato, K., Yoshimi, T., & Kondo, I. (2025). Validation of gait kinematics with ramp and stair ascent and descent revealed by markerless motion capture in simulated living space: Test-retest reliability study. JMIR Rehabilitation and Assistive Technologies, 12, e66886. https://doi.org/10.2196/66886
Link: Read the full study
Does Markerless Reliability Hold Up on Ramps and Stairs?
This study extends Theia3D's established day-to-day reliability beyond level walking to ramp and stair ascent and descent in a simulated living environment, reporting the first published reliability data for stair gait using markerless motion capture. If your work involves studying real-world mobility, assistive technology, or fall-prevention strategies across ramps and stairs, talk to our team about what this study could mean for your own research setting.