Exploring gait disorder characteristics in early Parkinson′s disease using artificial intelligence-assisted motor evaluation system
10.3760/cma.j.cn113694-20250507-00263
- VernacularTitle:利用人工智能运动评估系统探索早期帕金森病步态障碍特征
- Author:
Huijing LIU
1
;
Miaoxian XIE
;
Yueying LIU
;
Huimin CHEN
;
Wen SU
Author Information
1. 北京医院神经内科 国家老年医学中心 中国医学科学院老年医学研究院，北京 100730
- Publication Type:Journal Article
- Keywords:
Parkinson disease;
Gait;
Artificial intelligence;
Early diagnosis
- From:
Chinese Journal of Neurology
2025;58(9):938-945
- CountryChina
- Language:Chinese
-
Abstract:
Objective:To evaluate gait characteristics in early-stage Parkinson′s disease (PD) patients using an artificial intelligence-based quantitative motor function assessment system (Readygo) and validate whether PD patients with clinically normal gait actually exhibit objective gait impairments, and to explore the features and progression patterns of gait dysfunction in early PD.Methods:This cross-sectional, single-center study enrolled early-stage PD patients (Hoehn-Yahr stage≤2.5) from outpatient or inpatient departments of Beijing Hospital between October 2023 and October 2024, along with accompanying caregivers as healthy controls (HCs). Demographic data (sex, age, education level) were collected, and cognitive, psychological, and sleep-related scales assessments were administered. Based on the gait score (Item 3.10) from the Movement Disorder Society-Unified Parkinson′s Disease Rating Scale-Ⅲ (MDS-UPDRS-Ⅲ), PD patients were stratified into 3 subgroups: PD-normal gait (score=0), PD-mild gait impairment (score=1), and PD-moderate gait impairment (score=2). The Readygo system quantified gait parameters, including step width, stride length, step height, gait speed, stride velocity, swing velocity, and turn duration. Binary Logistic regression was uesd to identify biomarkers differentiating PD-normal gait group from HCs.Results:A total of 66 early-stage PD patients and 34 HCs were enrolled. Across the HCs, PD-normal gait, PD-mild gait impairment and PD-moderate gait impairment groups, there was a progressive decline in gait speed [1.07 (0.97, 1.15) m/s vs 0.97 (0.90, 1.06) m/s vs 0.90 (0.82, 1.00) m/s vs 0.77 (0.72, 0.86) m/s, H=29.949, P<0.001], bilateral stride velocity [left: 1.14 (1.07, 1.21) m/s vs 1.06 (0.94, 1.14) m/s vs 0.95 (0.88, 1.04) m/s vs 0.86 (0.76, 0.93) m/s, H=30.778, P<0.001; right: 1.12 (1.04, 1.22) m/s vs 1.04 (0.95, 1.13) m/s vs 0.96 (0.90, 1.04) m/s vs 0.89 (0.77, 0.90) m/s, H=29.561, P<0.001], and bilateral swing velocity [left: (2.56±0.28) m/s vs (2.38±0.32) m/s vs (2.19±0.33) m/s vs (1.96±0.32) m/s, F=14.132, P<0.001; right: 2.46 (2.35, 2.62) m/s vs 2.35 (2.13, 2.62) m/s vs 2.22 (2.05, 2.36) m/s vs 2.03 (1.71, 2.13) m/s, H=25.771, P<0.001], along with a progressive shortening of bilateral step length [left: 1.19 (1.14, 1.27) m vs 1.15 (1.04, 1.22) m vs 1.05 (0.93, 1.18) m vs 0.95 (0.80, 1.06) m, H=32.613, P<0.001; right: 1.20 (1.14, 1.30) m vs 1.13 (1.03, 1.22) m vs 1.07 (0.90, 1.17) m vs 0.97 (0.80, 1.03) m, H=30.528, P<0.001]. Conversely, turning time progressively lengthened [1.20 (1.09, 1.49) s vs 1.21 (1.10, 1.46) s vs 1.30 (1.19, 1.51) s vs 1.98 (1.53, 2.12) s, H=23.195, P<0.001]. Logistic regression identified that the right stride length was a discriminative factor between HCs and PD-normal gait group ( OR=0.023, 95% CI 0-0.291, P=0.012). Conclusions:As gait dysfunction worsens, PD patients demonstrate gradual reductions in speed-related parameters and stride length, with increasing turn duration.Early PD patients with clinically normal gait may already exhibit subtle impairments. Right stride length may serve as a potential biomarker to distinguish PD patients from HCs.
