Objective: To investigate gait and turning biomechanics in individuals with Parkinson’s disease using a 3D markerless motion capture system following a dry needling intervention.

Background: Individuals with Parkinson’s disease (PD) commonly present stooped posture and motor impairments such as rigidity and bradykinesia, which negatively affect gait performance. While dry needling may help alleviate muscle stiffness and improve mobility, quantitative gait assessment to capture these clinical changes often relies on laboratory-based motion capture systems that are costly and not widely accessible. Recent advances in artificial intelligence have enabled markerless motion capture approaches, offering a scalable alternative for objective gait analysis in clinical populations.

Method: Thirty-seven participants aged over 50 years, with a clinical diagnosis of PD (Hoehn and Yahr stages 1–3) and using levodopa, were evaluated at least one hour after their last dose. Participants were randomized into two groups (dry needling and sham). Gait was evaluated during a modified Timed Up and Go (TUG) test at baseline, immediately post-intervention, and at a 1-week follow-up. Video data were processed using the open-source vailá toolbox, which applied a spatial segmentation algorithm to isolate postural transitions, steady-state gait, and turning phases.

Results: Gait velocity showed a significant main effect of group (p < 0.001), with no effect of time (p=0.310), while the Group × Time interaction showed a trend toward significance (p=0.078). For the turning maneuver, turn time also showed a significant group effect (p < 0.001), with no effects of time (p=0.733) or interaction (p=0.384). Similarly, total TUG time showed only a significant group effect (p < 0.001), with no time (p=0.338) or interaction effects (p=0.507).

Conclusion: Group differences were observed for gait velocity, turning time, and total TUG time, while no significant changes across time were detected. These findings highlight the feasibility of markerless motion capture for detecting gait and turning characteristics in individuals with PD.

References: Bloem BR, Okun MS, Klein C. Parkinson’s disease. Lancet. 2021;397(10291):2284–2303.
Gupta R, Kumari S, Senapati A, Ambasta RK, Kumar P. New era of artificial intelligence and machine learning-based detection, diagnosis, and therapeutics in Parkinson’s disease. Ageing Res Rev. 2023;90:102013.
Tahara AK, Chinaglia AG, Monteiro RLM, Bedo BLS, Cesar GM, Santiago PRP. Predicting walkway spatiotemporal parameters using a markerless, pixel-based machine learning approach. Braz J Mot Behav. 2025;19(1):e462.
Vervullens S, Meert L, Baert I, Delrue N, Heusdens CHW, Hallemans A, Van Criekinge T, Smeets RJEM, De Meulemeester K. The effect of one dry needling session on pain, central pain processing, muscle co-contraction and gait characteristics in patients with knee osteoarthritis: a randomized controlled trial. Scand J Pain. 2022;22(2):396–409

« Back to 2026 International Congress

MDS Abstracts - https://www.mdsabstracts.org/abstract/markerless-kinematic-analysis-of-gait-and-turning-in-parkinsons-disease/