Gait Patterns and Balance Impairment in Parkinson’s Disease With Correlation to Disease Severity

Abstract

Background and Purpose

Accurate and early diagnosis of Parkinson’s disease (PD) remains challenging, as clinical assessments have notable limitations. Gait impairment, a hallmark motor symptom, appears in the early stages and is indicative of disease evolution. This study aimed to examine gait parameters and their associations with established clinical measures to assess their potential as diagnostic and monitoring biomarkers in a Korean PD cohort.

Methods

PD patients (n=18) and healthy controls (HCs, n=15) underwent gait assessments with MotionCore (JEIOS Inc.). Gait parameters were measured during both the 10-m walk (locomotion) and Timed Up and Go (TUG) tests, including stride length, cadence, single and double support time, toe-off time, and turning performance. Clinical assessments comprised the Hoehn & Yahr (H&Y) scale, Unified Parkinson’s Disease Rating Scale Part III, Freezing of Gait Questionnaire, Korean version of the Mini-Mental State Examination, Geriatric Depression Scale, and Berg Balance Scale.

Results

Patients with PD demonstrated significantly shortened stride length and higher cadence. Reduced single support time, extended double support time, and delayed toe-off time were observed in the PD group. Integrating gait metrics from both the 10-m locomotion and TUG tests improved discrimination between PD patients and healthy controls, as well as among different H&Y stages. Turning-related measures were significantly higher in PD patients. Gait measures showed robust correlations with clinical indices, including motor severity, cognitive status, and balance scores.

Conclusions

Gait analysis identified prominent motor and balance deficits in PD patients that were strongly associated with clinical severity. Gait parameters hold considerable potential as objective digital biomarkers for the diagnosis and monitoring of PD progression.

INTRODUCTION

Parkinson’s disease (PD) is the most prevalent movement disorder, affecting over 1% of adults older than 60 years.¹ Furthermore, PD represents the second most frequent neurodegenerative disease following Alzheimer’s disease (AD).²,³ The estimated global prevalence of individuals with PD is 25.2 million, with an anticipated increase of 76% across all age groups.³

The hallmark clinical symptoms of PD include rest tremors, rigidity, bradykinesia, and impaired postural reflexes.²,⁴ Beyond these motor manifestations, PD patients frequently present with non-motor symptoms, such as autonomic dysfunction, cognitive impairment, psychiatric disturbances, sensory alterations, and sleep disorders.² PD is mainly diagnosed based on clinical assessment, with neuroimaging studies performed to aid the diagnosis.¹ PD patients generally demonstrate reduced striatal 6-[18F]-fluoro-L-dopa uptake on positron emission tomography (PET).⁵ However, when clinical symptoms are detectable, 50%–70% of nigral dopaminergic neurons have already degenerated, underscoring the necessity for early detection of PD.⁶

Recently, there has been growing interest in digital biomarkers, particularly gait parameters, for the evaluation of PD patients.⁷ Gait impairment in PD, resulting from diminished postural reflexes, is a prominent motor symptom.⁴ Moreover, motor symptoms such as gait disturbances manifest in the early stages of PD and progressively worsen, significantly affecting daily functioning.⁸,⁹,¹⁰ Gait features in PD are characterized by asymmetric reduction of arm swing, slow turning, shuffling steps, increased temporal variability as the condition progresses, shortened step length, impaired gait rhythmicity, and freezing of gait.⁷ Although non-motor symptoms may precede motor manifestations, gait impairment gradually develops from the initial stage of PD, emphasizing the clinical relevance of evaluating gait patterns in PD.

Current evidence indicates that gait analysis is useful for tracking disease progression and assessing treatment outcomes in PD. Stride time is increased in PD patients and allows differentiation between healthy controls (HCs), early PD, and mild PD.¹¹,¹² In addition, gait performance has been shown to improve with dopaminergic therapy and deep brain stimulation surgery.¹³ Based on Hoehn and Yahr (H&Y) staging in PD, gait speed declines while cadence increases as the disease advances.¹⁴

Despite these findings, limited research has assessed quantitative gait parameters in PD patients within Korean populations.¹⁵,¹⁶ Furthermore, research examining the association between specific gait parameters and detailed clinical characteristics throughout the spectrum of PD severity remains scarce. Consequently, this study evaluated gait performance in a Korean cohort and explored its relationship with clinical variables. The incorporation of both the 10-m walk (locomotion) and Timed Up and Go (TUG) tests allows for a thorough evaluation of gait, assessing both gait dynamics and balance in individuals with PD.

METHODS

Patient collection

Patients diagnosed with PD and HCs were recruited from the Veterans Health Service Medical Center. PD patients were included if they fulfilled the UK Parkinson’s Disease Society Brain Bank criteria, as determined by a neurologist following at least six months of clinical monitoring, and were between 55 and 85 years old. The diagnosis of PD was confirmed using both brain MRI and 18F-N-(3-fluoropropyl)-2β-carboxymethoxy-3β-(4-iodophenyl) nortropane (FP-CIT) PET imaging. Age-matched HCs aged 55 to 85 years were also enrolled. HCs were required to be free from neurological, musculoskeletal, or other significant medical disorders that might influence gait. Health status for HCs was confirmed via self-reported medical history, hospital record review, and a brief physical assessment. Eligibility for HCs was based on the 28-item health screening guideline, ensuring the absence of known gait-impairing conditions. All participants were mandated to score within one standard deviation above the mean on the Korean version of the Mini-Mental State Examination (K-MMSE), relative to age- and education-adjusted norms, and to exhibit no motor symptoms. Demographic characteristics such as sex, height, and weight were recorded for all subjects. Motor evaluations included Unified Parkinson’s Disease Rating Scale (UPDRS), H&Y stage, Freezing of Gait Questionnaire (FOG-Q), Geriatric Depression Scale (GDS), and Berg Balance Scale (BBS), as well as cognitive (K-MMSE, Clinical Dementia Rating) and depression (Geriatric Depression Scale) assessments.

Informed consent was obtained from each participant or their legal guardian, and the study protocol received approval from the Institutional Review Board of the Veterans Healthcare Medical Center (2024-08-010). Written informed consent was obtained from the patient for publication.

Measurement of gait

Gait measurements were obtained using MotionCore (JEIOS Inc., Busan, Korea), which incorporates a camera-based visualization unit as well as sneakers embedded with gait sensors. Patients completed 2 types of gait assessments: locomotion and TUG. Prior to testing, each patient established their preferred walking speed with a metronome; this preferred speed served as the reference for determining slow and fast walking speeds. In the Locomotion test, patients walked 10-m 3 times at slow, normal, and fast walking speeds. For the TUG test, patients began seated on a chair without armrests, then, at the start of the measurement, stood up, walked 3-m, turned at a designated turning point, and returned to sit down. During the locomotion and TUG assessments, the following gait parameters were collected: stride length, cadence, single support, double support, time to toe, distance from start to turning onset (S), effective turning radius (ETR), and effective movement area (EMA).

Statistical analysis

Statistical analyses and graphical representations were conducted using GraphPad Prism software version 8.0.2 (GraphPad Software Inc., San Diego, CA, USA). Depending on the data distribution, either Student’s t-test or Mann-Whitney test was used for comparisons between 2 groups, and the p-value was calculated, with statistical significance set at p<0.05. The distribution of sex between groups was compared with Fisher’s exact test. Correlation analyses were performed using Spearman’s rank correlation coefficient. Data are presented as mean ± standard deviation.

RESULTS

Demographic characteristics of study participants

A total of 33 individuals participated in the study, comprising 18 patients with PD and 15 HCs (Table 1). The mean age of the PD group was 78 years (range, 48–84 years), while the mean age in the HC group was 72 years (range, 46–80 years), with no significant difference between the groups (p=0.1502). In the PD group, there were 16 males and 2 females; in the HC group, there were 13 males and 2 females, with no significant difference in sex distribution (p=0.8456). Height (p=0.3741) and weight (p=0.6359) did not differ significantly between groups.

Table 1. Demographic characteristics of study participants.

Characteristics		PD (n=18)	HC (n=15)	p-value
Age (yr)		77.61 (68–84)	72.07 (46–80)	0.1502*
Sex				>0.8456†
	Male	16 (88.9)	13 (86.7)
	Female	2 (11.1)	2 (13.3)
Height (cm)		164.89 (153–177)	166.47 (146–175)	0.3741*
Weight (kg)		64.17 (46–85)	66.53 (51–87)	0.6359*
H&Y stage				-
	1.0	6 (33.3)	-
	1.5	4 (22.2)	-
	2.0	8 (44.4)	-
UPDRS		57.50 (30–83)	-	-
K-MMSE		25.00 (17–30)	-	-
FOG-Q		10.28 (1–22)	-	-
GDS		3.20 (2–9)	-	-
BBS		39.94 (16–56)	-	-

Data are presented as number (%) or mean (range).

PD: Parkinson’s disease, HC: Healthy control, H&Y: Hoehn and Yahr, UPDRS: Unified Parkinson’s Disease Rating Scale, K-MMSE: Korean version of the Mini-Mental State Examination, FOG-Q: Freezing of Gait Questionnaire, GDS: Geriatric Depression Scale, BBS: Berg Balance Scale.

^*Mann Whitney test; ^†Fisher’s exact test.

Among the PD group, the mean H&Y stage included 6 patients at stage 1.0, 4 at stage 1.5, and 8 at stage 2.0. The mean UPDRS-III score was 58 (range, 30–83). The mean K-MMSE score was 25 (range, 17–30). Average scores for clinical assessments were as follows: FOG-Q, 10 (range, 1–22); GDS, 3 (range, 2–9); and BBS, 40 (range, 16–56).

Altered gait patterns in PD

PD patients demonstrated a significant reduction in stride length, particularly during the 10-m walk test (Fig. 1A and B). Stride length during the 10-m walk test was markedly decreased in patients with H&Y stages 1.5 and 2.0 compared to HC. In the TUG test, a reduction was also seen in H&Y stages 1.5 and 2.0, but statistical significance was only found in H&Y scale 1.5. Cadence was elevated in PD patients, showing an increasing trend in H&Y stages 1.0 and 1.5 but not in stage 2.0 (Fig. 1C and D). Velocity, determined by the product of stride length and cadence, was reduced in PD patients and exhibited a stepwise decline with higher H&Y stages (Fig. 2).

Fig. 1. Notable alterations in gait patterns among PD patients.

There was a significant reduction in stride length in both the 10-m walk (locomotion) and the TUG tests. (A) Stride length of the 10-m walk test were: HC: 0.935±0.0379, PD: 0.827±0.146; H&Y 1.0: 0.903±0.0914, H&Y 1.5: 0.789±0.110, H&Y 2.0: 0.789±0.182. (B) Stride length of the TUG test were: HC: 0.821±0.0843, PD: 0.738±0.146; H&Y 1.0: 0.792±0.126, H&Y 1.5: 0.700±0.0635, H&Y 2.0: 0.716±0.187. Cadence increased significantly in the 10-m walk and the TUG tests. (C) Evaluated cadence of the 10-m walk test were: HC: 106±5.95, PD: 108±12.9; H&Y 1.0: 114±9.33, H&Y 1.5: 113±11.8, H&Y 2.0: 103±14.4. (D) Evaluated cadence of the TUG test were: HC: 107±7.19, PD: 109±7.87; H&Y 1.0: 110±5.02, H&Y 1.5: 114±5.38, H&Y 2.0: 105±9.48.

Data are presented as mean ± standard deviation.

PD: Parkinson’s disease, TUG: Timed Up and Go, HC: healthy control, H&Y: Hoehn and Yahr.

^*p<0.05, ^**p<0.01.

Fig. 2. Reduced velocity (walking speed) in PD patients.

Changes in velocity were observed in both the 10-m walk (locomotion) and the TUG tests. (A) Changes in velocity of the 10-m walk test were: HC: 99.1±7.43, PD: 90.5±21.6; H&Y 1.0: 102±11.8, H&Y 1.5: 89.8±20.4, H&Y 2.0: 81.9±25.4. (B) Changes in velocity of the TUG test were: HC: 87.9±12.7, PD: 80.4±15.9; H&Y 1.0: 86.8±10.3, H&Y 1.5: 79.8±7.37, H&Y 2.0: 75.8±21.4.

Data are presented as mean ± standard deviation.

PD: Parkinson’s disease, TUG: Timed Up and Go, HC: healthy control, H&Y: Hoehn and Yahr.

Impairments in gait balance in PD

For assessment of gait balance, single support time, double support time, and Toe off time were measured. Single support time was shortened in PD patients, with greater reductions observed in the TUG test. In both the 10-m walk and TUG tests, H&Y stage 2.0 demonstrated significant decreases compared to HC, and significant differences were observed between H&Y stages 1.0 and 2.0 in the TUG test (Fig. 3A). Double support time was longer in PD patients, particularly in the TUG test. H&Y stage 2.0 displayed a marked increase compared to HC in both tests, as well as compared to H&Y stage 1.0 in the TUG (Fig. 3B). The time of Toe off was elevated in PD patients, with notable increases identified in the TUG test. H&Y stage 2.0 was significantly increased compared to HC in both the 10-m walk and TUG tests and was also higher than H&Y stage 1.0 in the TUG test (Fig. 3C).

Fig. 3. Gait balance impairments in PD patients.

(A) Single support time showed a significant reduction. 10-m walk test: HC: 41.0±1.71, PD: 40.1±2.17; H&Y 1.0: 41.3±1.59, H&Y 1.5: 41.3±1.50, H&Y 2.0: 38.7±2.04. TUG test: HC: 40.8±1.07, PD 39.2±2.45; H&Y 1.0: 40.7±1.38, H&Y 1.5: 40.2±1.42, H&Y 2.0: 37.6±2.61. (B) Double support time was markedly increased. 10-m walk test: HC: 18.1±3.45, PD: 19.7±4.39; H&Y 1.0: 17.3±3.24, H&Y 1.5: 17.4±3.16, H&Y 2.0: 22.7±4.06. TUG test: HC: 18.5±2.07, PD: 21.7±5.10; H&Y 1.0: 18.5±2.74, H&Y 1.5: 19.7±2.86, H&Y 2.0: 25.1±5.43. (C) Toe off time was significantly delayed. 10-m walk test: HC: 59.1±1.75, PD: 59.9±2.22; H&Y 1.0: 58.6±1.65, H&Y 1.5: 58.7±1.66, H&Y 2.0: 61.4±2.02. TUG test: HC: 59.3±1.02, PD: 60.9±2.66; H&Y 1.0: 59.2±1.37, H&Y 1.5: 59.8±1.47, H&Y 2.0: 62.8±2.83.

Data are presented as mean ± standard deviation.

PD: Parkinson’s disease, HC: healthy control, H&Y: Hoehn and Yahr, TUG: Timed Up and Go.

^*p<0.05, ^**p<0.01.

Enhanced differentiation by integrating gait parameters across tests

To enhance the differentiation between HC and PD, gait parameters from both the 10-m walk and TUG tests were analyzed in combination. Stride length was markedly reduced in PD participants, with those at H&Y stage 1.5 exhibiting significant decreases compared to HC (Fig. 4A). Cadence was found to be elevated in PD, with statistically significant increases at H&Y stages 1.0 and 1.5, whereas no significant change was detected at stage 2.0 (Fig. 4B). Single support time demonstrated a decline in PD, and was significantly shorter in H&Y stage 2.0 when compared with both HC and H&Y stage 1.0 (Fig. 4C). Double support duration was prolonged in individuals with PD, particularly at H&Y stage 2.0 which showed significant increases over HC and both H&Y stages 1.0 and 1.5 (Fig. 4D). The time to Toe off was longer in PD, with H&Y stage 2.0 exhibiting significant increases compared to HC, and H&Y stages 1.0 and 1.5 (Fig. 4E). Each set of combined gait parameters yielded a moderate classification performance between PD and HC, reflected by area under the curve (AUC) values ranging from 0.64 to 0.73 (Fig. 4F).

Fig. 4. Diagnostic significance of combining 10-m walk and TUG test according to H&Y stages.

(A) A significant reduction in stride length was observed. HC: 1.76±0.101, PD: 1.56±0.282; H&Y 1.0: 1.70±0.217, H&Y 1.5: 1.49±0.153, H&Y 2.0: 1.50±0.357. (B) Cadence was elevated in the PD and H&Y subgroups. HC: 213±11.8, PD: 217±19.8; H&Y 1.0: 224±13.1, H&Y 1.5: 227±16.4, H&Y 2.0: 208±23.0. (C) A marked decrease in single support time was detected. HC: 81.8±2.57, PD: 79.4±4.38; H&Y 1.0: 82.1±2.89, H&Y 1.5: 81.5±2.34, H&Y 2.0: 76.3±4.30. (D) Double support time was significantly increased. HC: 36.6±5.08, PD: 41.5±9.00; H&Y 1.0: 35.8±5.75, H&Y 1.5: 37.1±4.80, H&Y 2.0: 47.9±8.81. (E) A significant prolongation in the time of toe off was observed. HC: 118±2.53, PD: 121±4.63; H&Y 1.0: 118±2.87, H&Y 1.5: 119±2.49, H&Y 2.0: 124±4.52. (F) Receiver operating characteristic curve analyses assessed gait parameters as discriminative markers. Cadence: AUC=0.6407, specificity=80%, sensitivity=66.67%; Stride length: AUC=0.7296, specificity=100%, sensitivity=50%; Single support: AUC=0.6667, specificity=80%, sensitivity=55.56%; Double support: AUC=0.6667, specificity=86.67%, sensitivity=50%; Time of toe off: AUC=0.6556, specificity=86.67%, sensitivity=50%.

Data are presented as mean ± standard deviation.

TUG: Timed Up and Go, H&Y: Hoehn and Yahr, HC: healthy control, PD: Parkinson’s disease, AUC: area under the curve.

^*p<0.05, ^**p<0.01.

Turning performance as a significant aspect during TUG test

Turning-related metrics were recorded during the TUG test, including S, ETR, EMA, and total turning time. All metrics were substantially higher in PD participants, though the increase in S and decrease in speed did not reach statistical significance (Fig. 5). H&Y stage 2.0 was characterized by significant increases in S, EMR, and EMA compared with HC as well as H&Y stages 1.0 and 1.5 (Fig. 5A-C), but no significant difference in S was noted between H&Y stages 1.0 and 2.0.

Fig. 5. Significant increases in turning parameters in PD.

(A) Marked increase in the distance from start to turning onset was observed. HC: 2.60±0.436, PD: 2.81±0.448; H&Y 1.0: 2.72±0.303, H&Y 1.5: 2.35±0.356, H&Y 2.0; 3.10 ± 0.377. (B) Effective turning radius was significantly increased in PD and advanced H&Y stages. HC: 0.700±0.139, PD: 0.818±0.133; H&Y 1.0: 0.777±0.094, H&Y 1.5: 0.675±0.104, H&Y 2.0; 0.921±0.086. (C) The effective movement area was significantly greater in PD subjects and advanced H&Y stages. HC: 2.67±1.01, PD: 3.45±1.04; H&Y 1.0: 3.10±0.587, H&Y 1.5: 2.34±0.665, H&Y 2.0; 4.26±0.809. (D) Turning time was significantly prolonged in PD. HC: 8.85±0.971, PD: 10.9±3.48. (E) The velocity during turning was reduced in PD. HC: 0.798±0.084, PD: 0.739±0.118.

Data are presented as mean ± standard deviation.

PD: Parkinson’s, HC: healthy control, AUC: area under the curve disease, H&Y: Hoehn and Yahr.

^*p<0.05, ^**p<0.01.

Correlation between gait parameters and clinical characteristics

Spearman correlation analyses indicated that gait metrics were significantly associated with clinical indicators including H&Y stage, UPDRS, FOG-Q, K-MMSE, GDS, and BBS (Fig. 6). Stride length, single and double support time, and measures related to turning exhibited strong associations with the H&Y scale, UPDRS, and FOG-Q. Furthermore, notable correlations were found between gait performance and measures of cognitive function and balance.

Fig. 6. Correlations between gait parameters and clinical measures.

Correlation analyses were performed for (A) the 10-m walk test, (B) the TUG test, and (C) the combined assessment of both tests. The color scale depicts Spearman r values, and p-values are provided only for statistically significant correlations.

H&Y: Hoehn and Yahr, UPDRS-III: Unified Parkinson’s Disease Rating Scale Part III, FOG-Q: Freezing of Gait Questionnaire, K-MMSE: Korean version of the Mini-Mental State Examination, GDS: Geriatric Depression Scale, BBS: Berg Balance Scale, S: distance from start to turning onset, ETR: effective turning radius, EMA: effective movement area.

^*p<0.05, ^**p<0.01, ^***p<0.001, ^****p<0.0001.

DISCUSSION

Although the global prevalence of PD continues to rise, achieving early diagnosis remains a significant challenge.¹⁷ Dopaminergic neuronal loss disrupts the balance between the direct and indirect pathways in the striatum, contributing to the development of tremors, bradykinesia, and postural instability.¹⁸ However, objective quantification of these clinical features remains challenging. While the UPDRS and H&Y scale are used to assess motor impairment, their evaluations are based on clinician judgments, which limits objectivity in monitoring disease progression. Consequently, PD diagnosis remains heavily reliant on subjective clinical assessment, with a reported diagnostic accuracy of 80, underscoring the necessity for enhanced diagnostic approaches.¹⁹,²⁰,²¹

Gait dysfunction frequently presents as one of the earliest motor symptoms in PD.²²,²³ Recently, gait parameters have gained importance as diagnostic tools, largely because brain imaging is limited in its ability to detect PD.²⁴ These gait characteristics are being leveraged to develop quantitative methods for assessing motor function in PD.²⁵ Gait metrics such as stride length and cadence provide insight into PD motor manifestations, as reduced stride length often leads to compensatory increases in cadence to preserve walking speed despite shorter steps.²⁶ In the present study, we measured gait parameters in a Korean population and assessed their associations with clinical outcomes to validate their utility as digital biomarkers for PD. Gait analyses were conducted using both 10-m walk and TUG tests with MotionCore (JEIOS Inc.)

Stride length was markedly reduced in PD, whereas cadence increased as a compensatory response (Fig. 1). These patterns became increasingly evident with advancing H&Y stages, indicating that gait parameters reflect disease progression as documented in previous studies.²⁶ For the evaluation of gait balance, single support time, double support time, and toe off time were systematically examined. Among PD patients, there was a reduction in single support time accompanied by an increase in double support time, signifying a shift towards increased double support duration due to impaired balance (Fig. 3).²⁶ This compensatory adaptation was further evidenced by elevated cadence, as patients attempted to enhance stability by reducing the time spent on a single limb. These gait variables exhibited significant correlations with H&Y stage, with more pronounced associations evident in the TUG test. Toe off time was found to be delayed in PD patients and increased progressively with H&Y stage, representing challenges in initiating subsequent steps during ambulation. By combining assessments from the 10 m locomotion and TUG tests, the distinction between PD and HC, as well as differentiation across H&Y stages, was improved. Receiver operating characteristic analysis of gait metrics yielded moderate discriminative power, with AUC values between 0.64 and 0.73. Although specificity was relatively high (reaching 100% for stride length), sensitivity remained moderate (50%–66.67%) (Fig. 4). This result suggests enhanced ability to distinguish PD from HC, specifically by reducing false positive rates. Furthermore, assessment of turning parameters during the TUG test revealed significant increases in turning onset, effective turning radius, and effective movement area among PD patients (Fig. 5). Prior research established that anti-parkinsonian medications correlate with shorter turn durations and higher turning velocities in PD individuals.²⁷ Consistently, our results demonstrated relevant alterations in turning performance, supporting the utility of turning analysis as a valuable element in the comprehensive assessment of gait dysfunction in PD. Gait characteristics showed strong correlations with established clinical scales, including H&Y stage, UPDRS, and BBS, which are standard for evaluating motor dysfunction in PD (Fig. 6). In addition, gait parameters displayed inverse relationships with cognitive outcomes, underscoring the broader clinical relevance of gait analysis in PD.

However, this study has certain limitations. The sample size was limited, and there was an underrepresentation of female participants and controls compared to male participants. To mitigate this issue, future research should enroll larger and more balanced cohorts and utilize longitudinal designs to observe disease progression over time. Furthermore, cognitive impairment in PD involves multiple domains, and the MMSE applied in this study may lack sensitivity to detect executive dysfunction. The Montreal Cognitive Assessment may offer greater sensitivity for clinical relevance and should be considered in future investigations. Finally, although PD diagnosis in this cohort was supported by clinical ratings and 18F-FP-CIT PET, we did not perform a direct comparison between gait assessments and imaging outcomes. Incorporating neuroimaging findings could enhance understanding of the association between gait metrics and PD pathophysiology.

In summary, PD patients exhibited decreased stride length that was offset by an increase in cadence, along with impaired gait balance as evidenced by longer ground contact time and elevated double support time. Notably, gait instability resulted in longer turning time, as well as greater turning distance and area, features that robustly distinguished PD individuals from HC. Digital gait analysis offers objective, quantitative evaluation in clinical settings and may facilitate early detection of subtle motor deficits that appear prior to clear clinical or imaging evidence. Overall, these results reinforce the utility of gait characteristics as objective indicators for PD diagnosis and monitoring progression, and highlight their promise for development as digital biomarkers in clinical care.