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 not specified in the paper. This is a preprint (medRxiv), posted prior to peer review. The study assesses operational feasibility and workflow timing rather than kinematic accuracy or system agreement, and does not report comparative joint-angle or kinematic error data.
Study Overview
Article Type and Scope
A feasibility and operations-focused original research study (posted as a preprint, not yet peer-reviewed) evaluating whether high-throughput, portable markerless biomechanics data collection can be deployed consistently across on-campus and community-based sites, rather than a study of kinematic accuracy or system agreement.
Participants
Eighty-five adults aged 40 and older with a self-reported diagnosis of knee osteoarthritis in one or both knees, or who met American College of Rheumatology diagnostic criteria, recruited across four sites: a senior living community, a public senior recreation center, a university-affiliated laboratory, and a university student union. Participants were required to be able to walk 20 feet without assistance; exclusion criteria included a recent lower-limb trauma or surgery, recent joint replacement in the affected knee, and several other conditions affecting movement unrelated to osteoarthritis.
Data Collection
A portable markerless motion capture system, consisting of eight FLIR Blackfly S cameras and two Vicon Vero cameras (100 Hz) arranged in an oval pattern around a 3.6-meter walkway with two embedded force platforms, was transported by truck or hand cart to each site and processed using Theia3D software. Across twelve data collection events over an eleven-month period, participants completed informed consent, a set of six questionnaires, anthropometric measurement, and a motion capture assessment covering self-selected and fast-paced walking, double- and single-leg balance, squatting, sit-to-stand, and step-up/step-down tasks, with each activity's completion time recorded via synchronized timestamps.
Analysis
Timing metrics for research operations, including transit, setup, calibration, and breakdown, and for the participant-facing workflow, including informed consent, questionnaire completion, and motion capture, were compared across sites. Differences in participant-level timing were tested using Welch's one-way analysis of variance; research-operations timing differences were reported descriptively due to the limited number of data collection dates available per site.
Key Findings
- Participant-level timing showed no statistically significant differences across the four sites. Mean durations for informed consent (p = 0.495), questionnaires (p = 0.130), motion capture (p = 0.085), and total visit time (p = 0.946) were comparable across the senior living community, senior center, university laboratory, and student union sites, with total visit time varying by only about 2 minutes across sites on average.
- Average data collection time per participant was approximately 66 minutes, longer than the roughly 20 minutes per participant reported in an earlier, more limited community-based markerless feasibility study, a difference the authors attribute to this study's broader protocol: six questionnaires rather than two, a wider set of daily-living activities, and force-plate-embedded walking trials that required repeated attempts to capture clean kinetic data.
- Average idle time between study segments was low, at roughly 7 minutes per participant, despite higher enrollment and more walk-in participants at the community-based sites, reflecting a workflow in which study staff could flexibly reorder questionnaires and motion capture tasks in real time to avoid bottlenecks.
- Community-based sites attracted a higher proportion of walk-in and first-time research participants than on-campus sites, and overall site setup times trended longer at community-based locations than on campus (35.3 versus 26.6 minutes on average), though the limited number of data collection dates per site precluded formal statistical testing of this research-operations difference.
- Most participants completed the full activity battery: 69 of 85 participants (81.2%) completed every daily-living activity offered, with single-leg balance and step-up/step-down the activities most frequently skipped, and self-selected-pace walking trials showing the longest and most variable completion times (4.2 ± 1.7 minutes) of any activity, reflecting the repeated attempts often needed to register clean force-plate strikes.
What This Means for Movement Science Researchers and Lab Methodologists
For research teams weighing whether markerless motion capture can realistically extend biomechanics data collection beyond a single, fixed laboratory, this study offers a grounded, operations-first answer: a portable, multi-camera markerless system processed through Theia3D could be transported, set up, and run with consistent participant timing and low idle time across four very different locations, including two community-based sites outside a university setting. The finding that community sites drew more walk-in and first-time participants is a particularly useful signal for teams trying to build larger, more representative samples in conditions like osteoarthritis, where individual variability has long made small, single-site samples a limiting factor. At the same time, the modestly longer setup times observed at community sites, and the dependence of calibration efficiency on reliable remote technical support, are practical considerations worth planning around before scaling a similar model.
The Role of Theia3D in Movement Analysis
Theia3D was the software used to process video from the study's portable eight-camera FLIR and two-camera Vicon markerless array into kinematic and kinetic data, supporting data collection across all four sites in this feasibility study. Because this study's focus was operational feasibility and workflow timing rather than measurement accuracy, it does not report kinematic agreement data for Theia3D; its contribution here was as the processing engine underlying a high-throughput, portable data collection model deployed successfully in both on-campus and community settings.
Study Limitations and Future Directions
The authors note that the limited number of data collection dates available per site precluded formal statistical comparison of research-operations timing metrics such as setup, calibration, and breakdown, so those differences are reported descriptively rather than tested for significance. Interpretation of individual activity completion times was complicated by participant-specific variation in protocol, including participants who skipped activities entirely, attempted but did not complete certain tasks, or required more rest between activities than others, any of which could influence the recorded duration of neighboring tasks. Total visit time may modestly overestimate active data collection time for some participants, since the recorded end timestamp occasionally captured post-collection activities such as compensation processing or scheduling a follow-up visit rather than the final data collection task itself. The authors identify evaluating data quality and participant retention across sites, and optimizing study design elements such as team size and activity selection for specific research questions or populations, as clear next steps as this community-based model continues to be developed.
Citation: McCloskey, R. C., Qualter, J. M., Gruber, A., Leapley, S., Qiu, P., Tian, Z., Vincent, H. K., & Costello, K. E. (2026). Shared Strides: Operational feasibility of community-based biomechanics data collection in knee osteoarthritis. medRxiv. https://doi.org/10.64898/2026.04.20.26351135
Link: Read the full preprint
Can Markerless Capture Scale Beyond the Lab?
This study shows that a portable, Theia3D-processed markerless motion capture system can be deployed with consistent, efficient participant timing across both on-campus and community-based sites, a practical foundation for building larger, more representative biomechanical datasets. If your work involves extending motion capture data collection into community or field settings and you want to understand what that operational model could look like for your own research, talk to our team about deploying Theia3D beyond the traditional lab.