Objective: To identify and measure PD-specific gait signals in multi-camera videos for differentiating PD vs control (non-PD), without using specialized equipment (e.g., motion capture).

Background: Only 9% of individuals with PD are diagnosed and managed by movement disorder specialists (MDSs), with the majority under the care of general neurologists (GNs) and primary care physicians (PCPs) [1]. This research aims to make the detection of parkinsonian gait easier for GNs and PCPs to facilitate more accurate referrals to MDSs.

Method: We defined 13 a priori PD-specific gait signals based on clinical expertise of MDSs and published literature. We translated these clinical features into mathematical signal calculations based on 3D movement analysis of multi-camera videos of 10 PD and 5 non-PD subjects performing a 1-minute walk test. We analyzed gait-related feature scores across subjects and developed a predictive model to identify gait impairment.

Results: Compared to non-PD, PD subjects on average exhibited a 12-degree higher ankle flexion. During the 1-minute walk test, PD subjects showed a 2.3-fold reduction in walking velocity over the course of the test (progressively slowing down) while non-PD maintained their velocity. Additionally, PD subjects had 3-fold more asymmetry in arm swing, with a 2-fold decrease in arm swing amplitude when compared to non-PD. Elbow flexion asymmetry was also more pronounced in PD (1.43-fold asymmetry increase), with one elbow more flexed than the other. Turning duration was 5% longer in PD subjects than non-PD. PD subjects also showed a 44% reduction in gait cycle frequency and 12% shorter stride lengths. Using raw gait feature values as logistic regression classifier inputs, we predicted the presence of gait deficiencies with an 87% accuracy (compared to clinical consensus of 3 movement disorder specialists).

Conclusion: Preliminary findings show that PD-specific gait signals effectively distinguish PD gait patterns from non-PD patterns, offering a valuable tool for clinical decision-making in non-specialist settings. While many studies focus on differentiating PD from healthy controls, GNs and PCPs often face the challenge of determining if movement patterns are due to PD or another pathology. Future work will expand the study population to validate these features in distinguishing non-Parkinsonian gait issues, helping bridge the diagnostic gap for timely and appropriate treatment.

References: [1] Pearson C, Hartzman A, Munevar D, Feeney M, Dolhun R, Todaro V, Rosenfeld S, Willis A, Beck JC. Care access and utilization among medicare beneficiaries living with Parkinson’s disease. NPJ Parkinsons Dis. 2023 Jul 10;9(1):108. doi: 10.1038/s41531-023-00523-y. PMID: 37429849; PMCID: PMC10333279.

« Back to 2025 International Congress

MDS Abstracts - https://www.mdsabstracts.org/abstract/application-of-multi-camera-videography-for-detecting-pd-related-gait-deficiencies/