Quantitative assessment of gait characteristics in patients with Parkinson's disease using 2D video

Abstract

Introduction: Gait disorders are common in patients with Parkinson's disease (PD), and objective, quantifiable, and portable evaluation methods are needed in clinical practice. We assessed the differences in gait characteristics between patients with PD and healthy controls (HCs) using two-dimensional (2D) video and explored the diagnostic value of 2D video for early-stage PD.

Methods: Sixty-eight patients with PD and 48 HCs were recruited. All participants walked a standard 5-m track while recorded using mobile phones. The differences in gait characteristics between the groups and gait parameters of patients with early-stage PD were analyzed. Receiver operating characteristic analysis was used to evaluate the discrimination activity of gait parameters.

Results: Compared with HCs, patients with PD had shorter step length (S/M, p < 0.001/0.001), faster cadence (p = 0.044), slower walking speed (p < 0.001), smaller arm-swing angle (S/M, p < 0.001/0.001), slower arm-swing velocity (S/M, p < 0.001/0.001), and greater variability and asymmetry. The patients with early-stage PD had similar characteristics. After combined step length and variability and asymmetry of arm-swing angle, diagnostic sensitivity and specificity reached 84.6% and 89.6%, respectively, and an area under the curve of 0.91 (0.84-0.97).

Conclusion: We proposed a new method to improve the sensitivity of early-stage PD diagnosis using 2D video, which, as a portable and objective method, could be used to evaluate gait disorders in PD, including in early-stage PD. Combined diagnostic analysis of step length and variability and asymmetry of arm-swing angle could improve the sensitivity of PD diagnosis and was helpful in distinguishing patients with early-stage PD from HCs.

Keywords: 2D video; Early-stage Parkinson's disease; Gait disorder; OpenPose; Quantitative assessment.