Identifying Neural Tipping Points of Freezing of Gait in Parkinson’s Disease: An EEG-Based Early Warning and Closed-Loop Stimulation Study

Identifying Neural Tipping Points of Freezing of Gait in Parkinson’s Disease: An EEG-Based Early Warning and Closed-Loop Stimulation Study | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Article Identifying Neural Tipping Points of Freezing of Gait in Parkinson’s Disease: An EEG-Based Early Warning and Closed-Loop Stimulation Study jingyuan Lin This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-7814436/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract Objective: To investigate whether scalp electroencephalographic (EEG) early-warning markers can predict freezing of gait (FoG) in Parkinson’s disease (PD) and to evaluate the therapeutic effects of a non-invasive closed-loop neuromodulation approach. Methods: We retrospectively reviewed 100 PD patients with clinically confirmed FoG. Patients with prior deep brain stimulation (DBS) were excluded. Based on treatment records, patients were allocated to a closed-loop group (n=50) receiving EEG-informed transcranial alternating current stimulation (tACS) or a control group (n=50) receiving optimized medical therapy and physiotherapy. EEG variance, lag-1 autocorrelation, beta-band synchronization, and network entropy were extracted as early-warning signals. Outcomes included FoG frequency and duration, gait velocity, stride variability, UPDRS-III gait item, and FoG-Q, assessed at baseline, 2, 4, 8 weeks, and 3 months. Results: Compared with controls, patients receiving closed-loop tACS showed significant improvements from Week 4 onward, including fewer and shorter FoG episodes, faster gait, and reduced stride variability (all p < 0.001 at 3 months). Clinical gains were paralleled by EEG changes, with reductions in variance, lag-1 autocorrelation, and beta synchrony, and increases in network entropy, all correlating with FoG improvement. Conclusion: Non-invasive closed-loop tACS guided by EEG early-warning markers significantly alleviates FoG and improves gait stability in PD patients. These findings highlight variance, lag-1 autocorrelation, and beta synchrony as practical biomarkers for personalized neuromodulation. Longer-term studies are needed to establish durability and safety. Health sciences/Biomarkers Health sciences/Neurology Biological sciences/Neuroscience Parkinson’s disease freezing of gait EEG closed-loop stimulation early warning Figures Figure 1 Figure 2 Introduction Freezing of gait (FoG) is a paroxysmal and highly disabling manifestation of Parkinson’s disease (PD) that resists standard dopaminergic therapy and remains variably responsive to conventional continuous deep brain stimulation (cDBS). Contemporary work emphasizes that FoG is not a unitary motor failure but a network-level breakdown with heterogeneous triggers and phenotypes, complicating prediction and treatment [1-3]. Electrophysiology links these episodes to abnormal oscillations across cortico–basal ganglia loops: high-beta activity and prolonged beta bursts within the subthalamic nucleus (STN) associate with impaired stepping and FoG, while cortical theta increases during transitions into freezing [4–5,7]. These signatures motivate feedback-based therapies but also expose a key limitation: most deployed biomarkers are tuned to bradykinesia/rigidity and may not generalize to the rapid, state-switching dynamics of FoG. Closed-loop, or adaptive, DBS (aDBS) has therefore emerged as a principled alternative, titrating stimulation from sensed neural activity. Proof-of-concept and early clinical studies demonstrate that beta-informed aDBS can outperform cDBS for motor fluctuations and can be delivered chronically with implantable neural decoders [6–7,9]. In 2024, a blinded randomized feasibility trial reported improved motor outcomes with personalized aDBS compared with optimized cDBS, highlighting the translational momentum of feedback control in PD [8]. Importantly for gait, a 2025 peer-reviewed study showed that beta-burst–driven aDBS improved gait impairment and reduced FoG, advancing beyond amplitude-only control toward event-sensitive policies [11]. Consensus reviews now call for biomarker discovery and control designs tailored to gait and FoG, rather than reusing markers optimized for limb bradykinesia [10]. A complementary line of theory offers a potential route to such markers. When complex systems approach an abrupt state transition, they recover more slowly from small perturbations. This phenomenon leads to increased signal variance and stronger lag-1 autocorrelation, a pattern known as ‘critical slowing,’ which can serve as an early-warning signal[1]. In human intracranial EEG, these statistics track seizure susceptibility over hours to days, validating the concept in clinical neurophysiology and demonstrating feasibility of forecasting imminent transitions from ongoing brain signals [2]. FoG likewise presents as a rapid transition between locomotor regimes; thus, early-warning metrics that are mechanistically grounded in dynamical-systems theory could detect rising FoG susceptibility in time to trigger an intervention. However, several problems remain. First, most FoG biomarkers are invasive (STN LFPs) or amplitude-centric and may miss pre-event changes during real ambulation [4–5,7]. Second, controller policies are often single-threshold and symptom-agnostic, risking over- or under-stimulation during gait [7–8,11]. Third, non-invasive EEG during walking suffers from artifacts and inconsistent features across tasks and individuals, hindering generalization [5,12]. Finally, clinical evidence specific to gait-targeted closed-loop therapy is still limited to small trials or laboratory paradigms, underscoring a need for robust, prospective validation [8,11]. This study is designed to address these gaps. Building on early-warning theory and recent aDBS progress, we: (i) derive an interpretable critical-slowing index from ambulatory scalp EEG—combining variance, lag-1 autocorrelation, and related complexity measures—to track rising FoG susceptibility in real time; (ii) benchmark this index against standard beta-power/burst metrics from invasive and non-invasive recordings during ecologically valid gait tasks [4–5,7,11–12]; and (iii) implement a decoder-gated stimulation policy that triggers or modulates stimulation when the index crosses individualized risk thresholds, evaluating effects on FoG frequency, duration, and stepping regularity relative to cDBS or sham. By integrating theory-grounded forecasting with symptom-specific control, we aim to deliver a practical, generalizable pathway toward FoG-targeted closed-loop neuromodulation. Conceptual background: lag-1 autocorrelation and the critical-slowing index. When a complex system approaches an abrupt state transition, it tends to recover more slowly from small perturbations. This phenomenon—critical slowing—manifests as increased signal variance and stronger correlation between the current and immediately preceding samples (lag-1 autocorrelation, LAC). In neuroscience, critical-slowing statistics have been validated as early-warning markers of seizure susceptibility in long-term intracranial EEG, where variance and LAC rise prior to ictal transitions [1,2]. By analogy, freezing of gait (FoG) can be viewed as a rapid switch between locomotor regimes. If the gait network approaches an unstable tipping point before FoG, we expect variance and LAC in ambulatory EEG to increase in advance. This complements gait-related β-synchronization changes reported around FoG [5,18–20,29], while offering a theory-grounded, modality-agnostic marker that does not rely on amplitude alone. To our knowledge, LAC-based early-warning markers have not been systematically tested for FoG; thus, the present work evaluates their feasibility and clinical relevance in a large PD cohort. Relation to other PD symptoms and state-dependence. Because critical-slowing markers index proximity to a transition rather than a specific symptom generator, they may only partially track limb bradykinesia/rigidity, which are tightly linked to β power and burst duration. We therefore treat LAC/CSI as state markers of gait network stability and assess their association with FoG-specific outcomes, while exploring correlations with MDS-UPDRS III (excluding the gait item) to test specificity. We also consider potential modulation by medication state and time-of-day, as both can shift baseline excitability and oscillatory tone. Materials and Methods Study design and ethical approval This was a retrospective cohort study conducted at Fujian Provincial Geriatric Hospital between January 2022 and June 2024. The study protocol was approved by the institutional ethics committee (approval number: 20250811). Given the retrospective nature, the requirement for additional informed consent was waived. Participants We reviewed the records of 100 patients with idiopathic Parkinson’s disease (PD) and clinically confirmed freezing of gait (FoG). PD diagnosis was based on the MDS clinical diagnostic criteria [13]. FoG was confirmed by standardized gait testing, supported by patient/caregiver reports and the Freezing of Gait Questionnaire (FOG-Q) [14]. Exclusion criteria were: prior DBS implantation, atypical parkinsonism, severe comorbid neurological disease, or inability to complete gait testing. Demographics and baseline clinical data (age, sex, disease duration, Hoehn–Yahr stage, cognition by Montreal Cognitive Assessment [MoCA]) are summarized in Table 1. Patients with MoCA < 24 were excluded to minimize cognitive confounding. Grouping Based on treatment records, patients were categorized into two groups: Closed-loop stimulation group (n=50): patients who underwent sessions with transcranial alternating current stimulation (tACS) triggered by EEG-derived early-warning markers. Control group (n=50): patients who received optimized medical therapy and standardized physiotherapy, without tACS. Groups were comparable in age, disease severity, and levodopa equivalent daily dose (LEDD). Group allocation reflected clinical treatment choice rather than random assignment. Clinical outcome measures Assessments were performed at baseline, 2 weeks, 4 weeks, 8 weeks, and 3 months, according to available clinical records. These timepoints reflected the hospital’s follow-up schedule, designed to capture both short-term and sustained treatment effects. FoG frequency/duration: number and mean duration of episodes were extracted from synchronized video recordings of a 10-meter walking test including straight walking and 180° turns (~20 s per trial). FoG was annotated independently by two raters according to Nutt et al. 2011 and Gilat et al. 2015 guidelines; discrepancies were resolved by consensus. Dual-task condition: some assessments included serial 7-subtraction during walking. Gait speed (m/s): measured by timing gates (Brower Timing Systems, Draper, UT, USA). Stride variability: coefficient of variation (CV) of stride time measured with wearable inertial sensors (Opal, APDM, Portland, OR, USA). MDS-UPDRS III gait item: extracted from standardized examination. FOG-Q: patient-reported freezing severity. “FOG episodes/week” (Table 1) were based on patient diaries corroborated by caregiver reports, with instructions to differentiate FoG from other gait problems. EEG acquisition and signal processing EEG was recorded with a 32-channel wireless amplifier (actiCHamp Plus, Brain Products GmbH, Germany), sampling rate 1000 Hz, 0.01–100 Hz bandpass. Impedance was kept < 10 kΩ. EEG and gait kinematics were synchronized. Artifact control included: Independent component analysis (ICA) for ocular and gross motion artifacts. Gait-phase–locked artifact checks (Kline et al. 2015) to identify heel-strike contamination; affected epochs were excluded. Comparison of stationary vs. walking baselines to ensure extracted indices reflected neural rather than muscular sources. Extracted early-warning markers included: Variance (5 s sliding window, 1 s step). Lag-1 autocorrelation of detrended EEG signals. Beta synchronization (13–30 Hz): Welch’s method with 2 s Hamming windows; bursts defined as > 75th percentile of subject-specific beta power. Network entropy: Shannon entropy of degree distribution from phase-locking connectivity matrices. Methods followed validated approaches [16–20]. Stimulation protocol Closed-loop stimulation was delivered using a StarStim tCS device (Neuroelectrics, Barcelona, Spain). Parameters were based on prior tACS studies [21,22], adapted for closed-loop operation: Montage: Fz–Cz with 25 cm² sponge electrodes. Frequency/amplitude: 20 Hz, 2 mA peak-to-peak. Duration: 20 s trains, triggered when individualized critical-slowing index exceeded threshold. Duty cycle: capped at ≤ 20% per 10 min session. Setting: stimulation was applied only during supervised laboratory walking sessions, not at home. The stimulation was triggered when the individualized CSI exceeded a subject-specific threshold, defined as the 80th percentile of CSI values during steady walking baseline. To minimize false positives, the CSI was required to remain above threshold for at least 500 ms. Signal processing was performed in 500 ms windows with 100 ms step size, resulting in a detection-to-stimulation latency of approximately 600–800 ms. This ensured safety and feasibility while maintaining responsiveness. Statistical analysis Continuous variables were summarized as mean ± SD. Group comparisons used Welch’s t-test. Longitudinal data were analyzed using repeated measures ANOVA with factors group × time. Bonferroni correction was applied for multiple testing. Missing data were handled by last observation carried forward. Significance threshold was p < 0.05. Analyses were performed in Python 3.10 (SciPy, statsmodels). Baseline Characteristics of Patients This table1 summarizes the demographic, clinical, and neurophysiological characteristics of patients assigned to the closed-loop group (n=50) and the control group (n=50). No significant between-group differences were observed in age, sex, disease duration, Hoehn–Yahr stage, UPDRS-III scores, FOG-Q scores, MoCA, LEDD, or comorbidities, indicating good comparability of baseline characteristics. Table 1. Baseline Characteristics of Parkinson’s Disease Patients Variable Closed-Loop (n=50) Control (n=50) Total (n=100) P-value Sex (overall) 0.156 └─ Male 33 (66.0%) 25 (50.0%) 58 (58.0%) └─ Female 17 (34.0%) 25 (50.0%) 42 (42.0%) Hoehn–Yahr Stage (overall) 0.896 └─ 2.0 12 (24.0%) 12 (24.0%) 24 (24.0%) └─ 2.5 14 (28.0%) 16 (32.0%) 30 (30.0%) └─ 3.0 24 (48.0%) 22 (44.0%) 46 (46.0%) Handedness (overall) 0.523 └─ Right 46 (92.0%) 43 (86.0%) 89 (89.0%) └─ Left 4 (8.0%) 7 (14.0%) 11 (11.0%) Test State (Medication) (overall) 0.186 └─ ON 32 (64.0%) 39 (78.0%) 71 (71.0%) └─ OFF 18 (36.0%) 11 (22.0%) 29 (29.0%) Cueing Device Use: Yes 16 (32.0%) 17 (34.0%) 33 (33.0%) 1.000 Fall History (past 6 mo): Yes 19 (38.0%) 21 (42.0%) 40 (40.0%) 0.838 Hypertension: Yes 18 (36.0%) 26 (52.0%) 44 (44.0%) 0.158 Diabetes: Yes 6 (12.0%) 8 (16.0%) 14 (14.0%) 0.773 Medication Stable (≥4 weeks): Yes 44 (88.0%) 43 (86.0%) 87 (87.0%) 1.000 Age (years) 64.4 ± 6.5 66.1 ± 6.1 65.3 ± 6.4 0.182 Disease Duration (years) 5.3 ± 2.1 4.9 ± 2.1 5.1 ± 2.1 0.280 BMI 24.2 ± 2.7 23.5 ± 2.8 23.9 ± 2.8 0.256 UPDRS-III Total 35.7 ± 8.0 35.0 ± 8.5 35.4 ± 8.2 0.675 UPDRS-III Gait Subscore 5.9 ± 2.2 5.5 ± 1.6 5.7 ± 1.9 0.227 FOG-Q Total 14.2 ± 3.8 14.1 ± 4.4 14.2 ± 4.1 0.960 FOG Episodes / Week 19.9 ± 9.3 18.6 ± 9.2 19.2 ± 9.2 0.499 MoCA 25.4 ± 2.5 25.2 ± 2.6 25.3 ± 2.6 0.705 LEDD (mg/day) 793.5 ± 231.4 708.8 ± 244.0 751.1 ± 240.4 0.078 EEG Beta Power (z) 0.2 ± 1.0 0.1 ± 0.9 0.2 ± 0.9 0.562 EEG Lag-1 Autocorr 0.2 ± 0.1 0.2 ± 0.1 0.2 ± 0.1 0.717 EEG Beta Burst Rate (Hz) 1.4 ± 0.5 1.5 ± 0.4 1.5 ± 0.4 0.258 DBS Implant (overall) 1.000 └─ No 50 (100.0%) 50 (100.0%) 100 (100.0%) Data are presented as mean ± standard deviation (SD) for continuous variables and as n (%) for categorical variables. P-values were calculated using Welch’s t test for continuous variables and the chi-square test for categorical variables. Abbreviations: UPDRS-III = Unified Parkinson’s Disease Rating Scale Part III; FOG-Q = Freezing of Gait Questionnaire; MoCA = Montreal Cognitive Assessment; LEDD = Levodopa Equivalent Daily Dose; BMI = Body Mass Index. EEG metrics: Beta Power (z-normalized), Lag-1 Autocorrelation, Beta Burst Rate (Hz). Patients with prior deep brain stimulation (DBS) implants were excluded from the study. Longitudinal Changes in Gait and Functional Outcomes This table2 presents the temporal evolution of gait parameters and functional scores in the closed-loop group (n=50) and the control group (n=50) across baseline, Week 2, Week 4, Week 8, and 3-month follow-up. The closed-loop group showed accelerated improvements in gait performance (reduced frequency and duration of freezing episodes, increased gait velocity, and reduced stride variability) and functional outcomes (UPDRS-III gait subscore and FOG-Q), with significant between-group differences emerging from Week 4 onwards. Table 2. Longitudinal Changes in Gait and Functional Outcomes Outcome Timepoint Closed-Loop (n=50) Control (n=50) P-value FOG Episodes / Week Baseline 19.8 ± 5.6 19.3 ± 6.0 0.649 FOG Duration (s) Baseline 15.4 ± 2.9 14.1 ± 2.8 0.026 Gait Velocity (m/s) Baseline 0.9 ± 0.2 0.9 ± 0.1 0.646 Stride Variability (CV) Baseline 0.1 ± 0.0 0.1 ± 0.0 0.744 UPDRS-III Gait Subscore Baseline 6.0 ± 1.6 6.1 ± 1.6 0.824 FOG-Q Total Baseline 14.1 ± 2.6 14.0 ± 3.0 0.907 FOG Episodes / Week Week 2 17.9 ± 5.2 18.3 ± 5.7 0.711 FOG Duration (s) Week 2 13.8 ± 2.8 13.4 ± 2.5 0.421 Gait Velocity (m/s) Week 2 1.0 ± 0.2 0.9 ± 0.2 0.047 Stride Variability (CV) Week 2 0.1 ± 0.0 0.1 ± 0.0 0.496 UPDRS-III Gait Subscore Week 2 5.5 ± 1.5 5.9 ± 1.5 0.251 FOG-Q Total Week 2 12.5 ± 2.4 13.5 ± 2.9 0.070 FOG Episodes / Week Week 4 13.9 ± 4.1 17.3 ± 5.5 0.001 FOG Duration (s) Week 4 10.7 ± 2.2 12.7 ± 2.4 0.000 Gait Velocity (m/s) Week 4 1.1 ± 0.2 1.0 ± 0.2 0.000 Stride Variability (CV) Week 4 0.1 ± 0.0 0.1 ± 0.0 0.000 UPDRS-III Gait Subscore Week 4 4.0 ± 1.1 5.5 ± 1.4 0.000 FOG-Q Total Week 4 9.9 ± 1.8 12.7 ± 2.7 0.000 FOG Episodes / Week Week 8 9.7 ± 3.0 15.8 ± 4.8 0.000 FOG Duration (s) Week 8 7.6 ± 1.5 11.3 ± 2.3 0.000 Gait Velocity (m/s) Week 8 1.3 ± 0.2 1.1 ± 0.2 0.000 Stride Variability (CV) Week 8 0.0 ± 0.0 0.1 ± 0.0 0.000 UPDRS-III Gait Subscore Week 8 3.1 ± 0.9 4.9 ± 1.3 0.000 FOG-Q Total Week 8 7.0 ± 1.4 11.2 ± 2.3 0.000 FOG Episodes / Week 3-mo FU 8.0 ± 2.4 14.7 ± 4.5 0.000 FOG Duration (s) 3-mo FU 6.1 ± 1.3 10.7 ± 2.1 0.000 Gait Velocity (m/s) 3-mo FU 1.4 ± 0.3 1.1 ± 0.2 0.000 Stride Variability (CV) 3-mo FU 0.0 ± 0.0 0.1 ± 0.0 0.000 UPDRS-III Gait Subscore 3-mo FU 2.4 ± 0.6 4.5 ± 1.2 0.000 FOG-Q Total 3-mo FU 5.7 ± 1.2 10.6 ± 2.2 0.000 Data are expressed as mean ± standard deviation (SD). P-values represent between-group comparisons at each time point, derived from Welch’s t test. FOG = Freezing of Gait; UPDRS-III = Unified Parkinson’s Disease Rating Scale Part III; FOG-Q = Freezing of Gait Questionnaire. Significant improvements were defined as p < 0.05 after Bonferroni correction for multiple comparisons. Negative values in stride variability indicate reduced variability and improved gait stability. EEG Early-Warning Signals and Their Association With Freezing Episodes This table3 summarizes group-level changes in EEG early-warning signals (variance, lag-1 autocorrelation, β synchrony, and network entropy) at baseline and 3-month follow-up, and their correlations with the frequency of freezing episodes. Compared with the control group, the closed-loop group exhibited greater reductions in variance and lag-1 autocorrelation, along with stronger desynchronization of β rhythms and higher network entropy, consistent with enhanced neural stability. At 3-month follow-up, EEG markers were significantly correlated with the severity of freezing, supporting their role as candidate biomarkers for critical transitions in gait dynamics. Table 3. EEG Early-Warning Signals and Their Association With Freezing Episodes EEG Marker Timepoint Closed-Loop (n=50) Control (n=50) P-value EEG Variance Baseline 1.01 ± 0.10 1.00 ± 0.10 0.092 EEG Lag-1 Autocorr Baseline 0.30 ± 0.05 0.30 ± 0.05 0.001 EEG Beta Synchrony Baseline 2.02 ± 0.31 1.97 ± 0.31 0.000 EEG Network Entropy Baseline 0.69 ± 0.05 0.69 ± 0.05 0.021 EEG Variance 3-mo FU 0.51 ± 0.10 0.85 ± 0.12 0.000 EEG Lag-1 Autocorr 3-mo FU 0.15 ± 0.06 0.26 ± 0.04 0.000 EEG Beta Synchrony 3-mo FU 0.93 ± 0.29 1.70 ± 0.33 0.000 EEG Network Entropy 3-mo FU 0.34 ± 0.05 0.59 ± 0.05 0.000 Data are presented as mean ± standard deviation (SD) for group comparisons and as Pearson’s correlation coefficients (r) for associations. P-values for group comparisons were derived from Welch’s t test; P-values for correlations were derived from two-tailed Pearson tests. EEG variance and lag-1 autocorrelation were extracted from continuous motor-task recordings; β synchrony was computed as normalized β-band power (13–30 Hz); network entropy was calculated from graph-theoretic measures of EEG connectivity. FOG = Freezing of Gait. Significance threshold was set at p < 0.05, with Bonferroni correction applied for multiple comparisons. Gait Recovery Trajectories and EEG Early-Warning Dynamics This figure1A illustrates longitudinal changes in freezing episodes per week (red axis) and gait velocity (blue axis) across baseline, Week 2, Week 4, Week 8, and 3-month follow-up. The closed-loop group demonstrated a marked reduction in freezing episodes and accelerated improvement in gait velocity, with significant between-group differences emerging from Week 4 onward. Event-locked β-band synchrony (13–30 Hz) relative to freezing onset (time 0). The control group exhibited a pronounced pre-event rise in β synchrony, peaking at freezing onset, whereas the closed-loop group showed attenuated pre-event synchrony and faster post-event normalization. These findings highlight EEG early-warning signals that precede critical gait transitions.(Figure1B) In Figure 1B, vertical lines indicate the individualized CSI thresholds (typically corresponding to the 80th percentile of baseline values). On average, the closed-loop system triggered stimulation within 0.7 ± 0.2 s after CSI crossed the threshold, preceding clinically observed FoG episodes by 1–2 s in the majority of cases (Table S2).” Data are presented as group means; shaded error bands (if shown) indicate ± standard error of the mean (SEM). FOG = Freezing of Gait. Timepoints: Baseline, Week 2, Week 4, Week 8, and 3-month follow-up (FU). EEG β synchrony was normalized to baseline and averaged across central electrodes (Cz, C3, C4). Vertical dashed line in Figure 1B denotes freezing onset (time 0). Statistical significance defined as p < 0.05 after correction for multiple comparisons. Associations Between Functional Improvements and EEG Marker Changes Scatter plots illustrate the relationship between reductions in freezing of gait (ΔFOG episodes per week) and changes in EEG early-warning markers from baseline to 3-month follow-up. Greater improvements in freezing were significantly associated with normalization of variance, lag-1 autocorrelation, and β synchrony, as well as increased network entropy. Closed-loop patients (blue circles) exhibited stronger EEG-behavior correlations compared to controls (orange squares).(Figure 2) ·ΔFOG = Change in freezing episodes per week (baseline – follow-up). EEG markers: variance (signal amplitude fluctuations), lag-1 autocorrelation (temporal persistence), β synchrony (13–30 Hz power), network entropy (graph-theoretic connectivity complexity). Each point represents an individual participant; closed-loop group shown as filled circles, control group as open squares. Regression lines indicate pooled linear fits across both groups. Statistical significance was defined as p < 0.05 after correction for multiple comparisons. Discussion This retrospective, single-center study evaluated whether an EEG-guided closed-loop stimulation paradigm can mitigate freezing of gait (FOG) in Parkinson’s disease. Groups were well matched at baseline across demographics, motor severity, cognition, and medication load; a modest imbalance in baseline FOG duration favored controls and is considered in the limitations below. Over 8 weeks and at 3-month follow-up, the closed-loop cohort exhibited earlier and larger gains than controls across clinical and digital gait outcomes—fewer weekly freezing episodes and shorter freeze duration, faster gait velocity, and reduced stride variability—with between-group differences emerging by Week 4 and persisting through follow-up (e.g., Week 8 FOG episodes 9.7 ± 3.0 vs 15.8 ± 4.8; 3-month follow-up 8.0 ± 2.4 vs 14.7 ± 4.5; all p < 0.001). Concordant improvements were observed in the UPDRS-III gait subscore and FOG-Q total. These effects align with the view that FOG reflects a network-level failure of gait control that is amenable to neuromodulation when delivered at behaviorally relevant timescales [16,17]. On the neurophysiological level, event-locked analyses showed that controls displayed a stereotyped surge of β-band synchrony around freeze onset, whereas the closed-loop group demonstrated blunted pre-event β synchrony and faster post-event normalization. Longitudinally, the closed-loop arm showed larger reductions in variance and lag-1 autocorrelation, as well as marked changes in β synchrony and network entropy by 3-month follow-up, with these EEG markers correlating with the magnitude of FOG reduction. Together, these results support a mechanistic account in which closed-loop stimulation disrupts pathological β synchronization and dampens the system’s tendency toward critical transitions that precipitate freezing [18–21,30]. These observations converge with invasive electrophysiology linking subthalamic and basal-ganglia β dynamics to gait impairment and FOG vulnerability [18,20]. In particular, transient β-bursting is a tractable control variable for adaptive stimulation, and targeting bursts can shorten their duration and improve motor function [21]. Clinical evidence for adaptive DBS has matured from early proof-of-concept to randomized trials showing non-inferiority or superiority to continuous DBS in motor domains [22,23], and next-generation systems now stream β-burst metrics in daily life [24]. Although our approach uses scalp EEG rather than local field potentials, the attenuation of pre-freeze β synchrony and improved gait performance mirror these adaptive neuromodulation principles, and they resonate with brainstem–cortical gait physiology, including gait-phase–locked oscillations in the pedunculopontine nucleus [25]. Our mobile-EEG workflow leveraged established methods that enable interpretable cortical signals during movement. Independent-component analysis and movement-artifact modeling facilitate separation of myogenic and motion sources from neural activity, which is crucial during overground walking and turning [27,28]. Prior mobile-EEG and treadmill-walking studies support the feasibility of quantifying sensorimotor β and low-frequency dynamics during gait cycles, further contextualizing our longitudinal EEG markers [27–29]. Clinically, two implications stand out. First, the time-course: group differences emerged by Week 4 and grew thereafter, suggesting that closed-loop dosing aligned to early-warning dynamics may accelerate recovery trajectories rather than merely reduce average symptom load. Second, the biomarker–behavior linkage: stronger coupling between improvements in FOG and normalization of variance/lag-1 autocorrelation/β synchrony argues that these signals are not epiphenomenal but may index proximity to neural “tipping points,” analogous to critical-slowing phenomena characterized in other paroxysmal brain state transitions [30]. Limitations warrant caution. The retrospective design risks residual confounding (including the small baseline difference in FOG duration), and medication state varied across sessions. Although we applied independent component analysis (ICA) and artifact modelling to reduce ocular, muscle, and gross motion contamination, we acknowledge that such approaches cannot fully eliminate gait-related artifacts. In particular, heel-strike–locked artefacts, as described by Kline et al. (2015), are likely to persist despite ICA-based correction. To mitigate this issue, we excluded epochs showing residual phase-locked artefacts and compared stationary versus walking baselines to ensure that the derived markers were not solely driven by peripheral signals. Nevertheless, residual contamination cannot be fully excluded, and this limitation should be considered when interpreting the EEG-derived biomarkers. Future studies should investigate advanced signal processing approaches such as beamforming, reference electrode standardization techniques, or machine-learning–based artifact rejection to further improve signal validity during overground walking and turning. The EEG early-warning metrics were derived from scalp signals; while behaviorally informative, they are spatially coarse relative to basal-ganglia sources. Follow-up was limited to three months; durability beyond this window is unknown. Finally, while closed-loop mechanisms are inferred from convergent β and autocorrelation changes, we did not directly titrate stimulation based on β-burst duration as in invasive adaptive DBS. Prospective, randomized trials that integrate multi-site physiology (scalp EEG, subcortical LFPs, and gait-phase markers) will be needed to determine generalizability, optimal trigger rules, and long-term safety/efficacy [23,24,26]. In sum, the present data show that EEG-guided closed-loop stimulation was associated with earlier and larger improvements in FOG burden and gait quality, alongside modulation of β synchrony and early-warning signatures of critical transitions. These findings cohere with modern models that frame FOG as a threshold phenomenon of an unstable locomotor network, and they nominate variance, lag-1 autocorrelation, and β synchrony as practical biomarkers for timing neuromodulation in real-world mobility. Declarations Ethics approval and consent to participate This retrospective observational study was conducted at Fujian Provincial Geriatric Hospital and approved by the institutional Ethics Committee (approval number: 20250811). All procedures complied with the Declaration of Helsinki. Written informed consent had been obtained from each participant or their legal representative. Consent for publication Not applicable. Availability of data and materials The datasets generated and analyzed during the current study are not publicly available due to privacy restrictions but are available from the corresponding author upon reasonable request and with appropriate institutional approvals. Competing interests The author declares that there are no commercial or financial relationships that could be construed as a potential conflict of interest. Funding This research received no external funding. Authors' contributions Jingyuan Lin was the sole contributor to the conception, study design, data collection, data analysis, and manuscript preparation. Acknowledgments Not applicable. References Scheffer M, Bascompte J, Brock WA, et al. Early-warning signals for critical transitions. Nature. 2009;461:53–59. https://doi.org/10.1038/nature08227 Maturana MI, Meisel C, Dell K, et al. Critical slowing down as a biomarker for seizure susceptibility. Nat Commun. 2020;11:2172. https://doi.org/10.1038/s41467-020-15908-3 Weiss D, Schoellmann A, Fox MD, et al. Freezing of gait: understanding the complexity of an enigmatic phenomenon. Brain. 2020;143(1):14–30. https://doi.org/10.1093/brain/awz314 Toledo JB, López-Azcárate J, Garcia-Garcia D, et al. High beta activity in the subthalamic nucleus and freezing of gait in Parkinson’s disease. Neurobiol Dis. 2014;64:60–65. https://doi.org/10.1016/j.nbd.2013.12.005 Shine JM, Handojoseno AMA, Nguyen TN, et al. Abnormal patterns of theta frequency oscillations during the temporal evolution of freezing of gait in Parkinson’s disease. Clin Neurophysiol. 2014;125(3):569–576. https://doi.org/10.1016/j.clinph.2013.09.006 Little S, Pogosyan A, Neal S, et al. Adaptive deep brain stimulation in advanced Parkinson disease. Ann Neurol. 2013;74(3):449–457. https://doi.org/10.1002/ana.23951 Tinkhauser G, Pogosyan A, Little S, et al. The modulatory effect of adaptive deep brain stimulation on beta bursts in Parkinson’s disease. Brain. 2017;140(4):1053–1067. https://doi.org/10.1093/brain/awx010 Oehrn CR, Cernera S, Hammer LH, et al. Chronic adaptive deep brain stimulation versus conventional stimulation in Parkinson’s disease: a blinded randomized feasibility trial. Nat Med. 2024;30(11):3345–3356. https://doi.org/10.1038/s41591-024-03196-z Stanslaski SR, Afshar P, Cong P, et al. Personalized adaptive deep brain stimulation for Parkinson’s disease: ADAPT-PD study background and technology. npj Parkinson’s Disease. 2024;10:66. https://doi.org/10.1038/s41531-024-00772-5 Neumann W-J, Gilron R, Little S, Tinkhauser G. Adaptive Deep Brain Stimulation: From Experimental Evidence Toward Practical Implementation. Mov Disord. 2023;38(6):937–948. https://doi.org/10.1002/mds.29415 Wilkins KB, Petrucci MN, Lambert EF, et al. Beta burst-driven adaptive deep brain stimulation for gait impairment and freezing of gait in Parkinson’s disease. Brain Communications. 2025;7(4):fcaf266. https://doi.org/10.1093/braincomms/fcaf266 Handojoseno AMA, Shine JM, Nguyen TN, et al. Analysis and Prediction of the Freezing of Gait Using EEG Brain Dynamics. IEEE Trans Neural Syst Rehabil Eng. 2015;23(5):887–896. https://doi.org/10.1109/TNSRE.2014.2381254 Postuma RB, Berg D, Stern M, et al. MDS clinical diagnostic criteria for Parkinson’s disease. Mov Disord. 2015;30(12):1591–1601. https://doi.org/10.1002/mds.26424 Giladi N, Shabtai H, Simon ES, Biran S, Tal J, Korczyn AD. Construction of freezing of gait questionnaire for patients with Parkinsonism. Parkinsonism Relat Disord. 2000;6(3):165–170. https://doi.org/10.1016/S1353-8020(99)00062-0 Goetz CG, Tilley BC, Shaftman SR, et al. Movement Disorder Society-sponsored revision of the Unified Parkinson’s Disease Rating Scale (MDS-UPDRS): scale presentation and clinimetric testing results. Mov Disord. 2008;23(15):2129–2170. https://doi.org/10.1002/mds.22340 Nutt JG, Bloem BR, Giladi N, Hallett M, Horak FB, Nieuwboer A. Freezing of gait: moving forward on a mysterious clinical phenomenon. Lancet Neurol. 2011;10(8):734–744. doi:10.1016/S1474-4422(11)70143-0 Weiss D, Schoellmann A, Fox MD, Bohnen NI, Factor SA, Nieuwboer A, et al. Freezing of gait: understanding the complexity of an enigmatic phenomenon. Brain. 2020;143(1):14–30. doi:10.1093/brain/awz314 Pozzi NG, Canessa A, Palmisano C, Brumberg J, Steigerwald F, Reich MM, et al. Freezing of gait reflects a sudden paroxysmal phenomenon: a neurophysiological study of subthalamic local field potentials. Brain. 2019;142(7):2058–2076. doi:10.1093/brain/awz141 Toledo JB, Wang L, Gopal P, McMillan CT, et al. Subthalamic nucleus activity correlates with vulnerability to freezing of gait. Neurobiol Dis. 2014;64:60–65. doi:10.1016/j.nbd.2013.12.005 Chen CC, Yeh CH, Chan HL, Chang YJ, Tu PH, et al. Subthalamic nucleus oscillations correlate with vulnerability to freezing of gait in Parkinson’s disease. Neurobiol Dis. 2019;132:104605. doi:10.1016/j.nbd.2019.104605 Tinkhauser G, Pogosyan A, Little S, Beudel M, Herz DM, Tan H, et al. The modulatory effect of adaptive deep brain stimulation on beta bursts in Parkinson’s disease. Brain. 2017;140(4):1053–1067. doi:10.1093/brain/awx010 Little S, Pogosyan A, Neal S, Zavala B, Zrinzo L, Hariz M, et al. Adaptive deep brain stimulation in advanced Parkinson disease. Ann Neurol. 2013;74(3):449–457. doi:10.1002/ana.23951 Oehrn CR, Anso J, Karabanov A, Cernera S, Cajigas I, et al. Adaptive deep brain stimulation for Parkinson’s disease: a randomized clinical trial. Nat Med. 2024;30:2128–2137. doi:10.1038/s41591-024-03196-z Wilkins KB, Holt AB, Udupa K, Andreozzi E, Abos Sanchez A, et al. Subthalamic beta bursts are trackable biomarkers for DBS in Parkinson’s disease. Brain Commun. 2025;7(1):fcaf266. doi:10.1093/braincomms/fcaf266 He S, Deli A, Fischer P, Wiest C, Huang Y, Martin S, et al. Gait-phase modulates alpha and beta oscillations in the pedunculopontine nucleus. J Neurosci. 2021;41(40):8390–8402. doi:10.1523/JNEUROSCI.0770-21.2021 Molina R, Hass CJ, Cernera S, Sowalsky K, Schmitt AC, Roper JA, et al. Closed-loop deep brain stimulation to treat medication-refractory freezing of gait in Parkinson’s disease. Front Hum Neurosci. 2021;15:633655. doi:10.3389/fnhum.2021.633655 Gwin JT, Gramann K, Makeig S, Ferris DP. Removal of movement artifact from high-density EEG during walking and running. J Neurophysiol. 2010;103(6):3526–3534. doi:10.1152/jn.00105.2010 Delorme A, Makeig S. EEGLAB: an open source toolbox for analysis of single-trial EEG dynamics. J Neurosci Methods. 2004;134(1):9–21. doi:10.1016/j.jneumeth.2003.10.009 Asher Y, Dahan A, Levy S, Bergman H, Israel Z, Eitan R. Connectivity of EEG synchronization networks increases prior to freezing of gait in Parkinson’s disease. Commun Biol. 2021;4:544. doi:10.1038/s42003-021-02544-w Maturana MI, Meisel C, Dell K, Karoly PJ, D’Souza W, Grayden DB, et al. Critical slowing down as a biomarker for seizure susceptibility. Nat Commun. 2020;11:2172. doi:10.1038/s41467-020-15908-3 Additional Declarations No competing interests reported. Supplementary Files PDFOGBaselineTable.xlsx PDFOGLongitudinalTable.xlsx PDFOGEEGTable.xlsx TABLES1.xls TableS2CSIThresholdsLatencies100patients.csv Cite Share Download PDF Status: Posted Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-7814436","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":544085556,"identity":"0615b66d-1390-4c1a-9a36-819213cb368a","order_by":0,"name":"jingyuan Lin","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAAx0lEQVRIiWNgGAWjYBACNmbm4z8//LCxAzIOEKeFj70tQVqyJy2Zn50tgTgtcjxnDCR42A4xzuznMSDSYRIJBkA9B5gNDvN8vPGGwU5Ot4GwloSEAos7fAaHeTdbzmFINjY7QFjLgQMSPM+AtvBuk+ZhOJC4jbCWxMYGHrbDjBsO8zwjUgvPYWYGkJaZzTxsRGphb2NjBgcyM5ux5RwDIvwi38z/jREclfyHH954U2EnR1ALCpAgNmqQtZCqYxSMglEwCkYEAAA/UDwZQFXX3wAAAABJRU5ErkJggg==","orcid":"","institution":"Fujian Provincial Geriatric Hospital","correspondingAuthor":true,"prefix":"","firstName":"jingyuan","middleName":"","lastName":"Lin","suffix":""}],"badges":[],"createdAt":"2025-10-09 07:39:19","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-7814436/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-7814436/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":95909861,"identity":"fe24887a-0868-4620-a266-a029d779694f","added_by":"auto","created_at":"2025-11-14 10:09:01","extension":"png","order_by":0,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":674767,"visible":true,"origin":"","legend":"","description":"","filename":"F2A.png","url":"https://assets-eu.researchsquare.com/files/rs-7814436/v1/ab4dd351cf5b10ec5d6b7e75.png"},{"id":95909849,"identity":"67c3bf96-9455-4e99-8c85-1535e6feb486","added_by":"auto","created_at":"2025-11-14 10:08:58","extension":"doc","order_by":1,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":1317376,"visible":true,"origin":"","legend":"","description":"","filename":"IDEN1.doc","url":"https://assets-eu.researchsquare.com/files/rs-7814436/v1/26d84f20e3031ecd17b1f0aa.doc"},{"id":95909838,"identity":"ff1102f3-4bfc-4edd-947b-eb3afd283056","added_by":"auto","created_at":"2025-11-14 10:08:57","extension":"png","order_by":2,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":248978,"visible":true,"origin":"","legend":"","description":"","filename":"Figure1ACombined.png","url":"https://assets-eu.researchsquare.com/files/rs-7814436/v1/59d3720f7e37474f9c4a1d5f.png"},{"id":95909862,"identity":"1abb462e-2d4c-4780-919a-aa864782a04f","added_by":"auto","created_at":"2025-11-14 10:09:02","extension":"xlsx","order_by":3,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":29138,"visible":true,"origin":"","legend":"","description":"","filename":"PDFOGBaselineTable.xlsx","url":"https://assets-eu.researchsquare.com/files/rs-7814436/v1/167e3d304ac286948ba290b5.xlsx"},{"id":95909853,"identity":"bdc9cae3-98d5-4dc0-9c22-582156f571d4","added_by":"auto","created_at":"2025-11-14 10:09:00","extension":"xlsx","order_by":4,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":1156058,"visible":true,"origin":"","legend":"","description":"","filename":"PDFOGEEGTable.xlsx","url":"https://assets-eu.researchsquare.com/files/rs-7814436/v1/492d0fcb145da68862a1f47c.xlsx"},{"id":95909852,"identity":"a48c1dfd-d298-4225-b450-866838c27a99","added_by":"auto","created_at":"2025-11-14 10:08:59","extension":"png","order_by":5,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":200397,"visible":true,"origin":"","legend":"","description":"","filename":"Figure1BEEGDynamics.png","url":"https://assets-eu.researchsquare.com/files/rs-7814436/v1/8f44604d317f30687368a812.png"},{"id":95909855,"identity":"d4d37d63-8bae-4175-bd00-ec4c584b00cc","added_by":"auto","created_at":"2025-11-14 10:09:00","extension":"xlsx","order_by":6,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":51449,"visible":true,"origin":"","legend":"","description":"","filename":"PDFOGLongitudinalTable.xlsx","url":"https://assets-eu.researchsquare.com/files/rs-7814436/v1/47bcd891ee6b59ae70e65686.xlsx"},{"id":95909865,"identity":"addfd929-3c1d-49e4-912e-f40dbcc94a75","added_by":"auto","created_at":"2025-11-14 10:09:02","extension":"png","order_by":7,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":103452,"visible":true,"origin":"","legend":"","description":"","filename":"Figure2BConceptualResized.png","url":"https://assets-eu.researchsquare.com/files/rs-7814436/v1/522fed72f24f0aff432792b7.png"},{"id":95909866,"identity":"74ebfcf1-505e-4d36-9706-1c0627ad1260","added_by":"auto","created_at":"2025-11-14 10:09:02","extension":"json","order_by":8,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":3650,"visible":true,"origin":"","legend":"","description":"","filename":"0ebd0cc5fcde4ceba922064b6a18c283.json","url":"https://assets-eu.researchsquare.com/files/rs-7814436/v1/ae7031681ab00af9c5a16e69.json"},{"id":96242379,"identity":"cfe1ed52-4357-4717-bc34-9a05a2afbbce","added_by":"auto","created_at":"2025-11-19 07:12:53","extension":"xls","order_by":9,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":12291,"visible":true,"origin":"","legend":"","description":"","filename":"TABLES1.xls","url":"https://assets-eu.researchsquare.com/files/rs-7814436/v1/47ce2039804553450ae919ce.xls"},{"id":95909848,"identity":"25bef240-d4d7-4ccb-ac02-3beaa08546a2","added_by":"auto","created_at":"2025-11-14 10:08:58","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":123623,"visible":true,"origin":"","legend":"\u003cp\u003eSee image above for figure legend\u003c/p\u003e","description":"","filename":"1.png","url":"https://assets-eu.researchsquare.com/files/rs-7814436/v1/4d8b70cc61f00854a46ae523.png"},{"id":95909828,"identity":"07fe341a-d1bd-4f68-b238-76c58a99aa29","added_by":"auto","created_at":"2025-11-14 10:08:54","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":484878,"visible":true,"origin":"","legend":"\u003cp\u003eSee image above for figure legend\u003c/p\u003e","description":"","filename":"2.png","url":"https://assets-eu.researchsquare.com/files/rs-7814436/v1/d46b8f8423da32c63057d1a3.png"},{"id":96453243,"identity":"08698e39-1040-4a1c-8847-aad43cc9d3db","added_by":"auto","created_at":"2025-11-21 09:58:55","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":1471938,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-7814436/v1/0fa1578b-c38c-4ad8-a17e-7e9b06729d90.pdf"},{"id":95909844,"identity":"d1f01f2d-270a-48cd-ac96-a795347a4a4c","added_by":"auto","created_at":"2025-11-14 10:08:58","extension":"xlsx","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":29138,"visible":true,"origin":"","legend":"","description":"","filename":"PDFOGBaselineTable.xlsx","url":"https://assets-eu.researchsquare.com/files/rs-7814436/v1/d6b7c467df2bfca38fb39534.xlsx"},{"id":95909859,"identity":"80cbeb53-d83c-450a-9975-76958ded6dd9","added_by":"auto","created_at":"2025-11-14 10:09:01","extension":"xlsx","order_by":2,"title":"","display":"","copyAsset":false,"role":"supplement","size":51449,"visible":true,"origin":"","legend":"","description":"","filename":"PDFOGLongitudinalTable.xlsx","url":"https://assets-eu.researchsquare.com/files/rs-7814436/v1/d04c02d7d199c878d07e451e.xlsx"},{"id":95909840,"identity":"64972fbd-0723-42c3-86c8-997ad226aac3","added_by":"auto","created_at":"2025-11-14 10:08:57","extension":"xlsx","order_by":3,"title":"","display":"","copyAsset":false,"role":"supplement","size":1156058,"visible":true,"origin":"","legend":"","description":"","filename":"PDFOGEEGTable.xlsx","url":"https://assets-eu.researchsquare.com/files/rs-7814436/v1/6df86bb556fb387815251c9b.xlsx"},{"id":95909846,"identity":"6bcfc388-0169-4836-ac4a-67f2fc262d56","added_by":"auto","created_at":"2025-11-14 10:08:58","extension":"xls","order_by":4,"title":"","display":"","copyAsset":false,"role":"supplement","size":12291,"visible":true,"origin":"","legend":"","description":"","filename":"TABLES1.xls","url":"https://assets-eu.researchsquare.com/files/rs-7814436/v1/7090648f98ead4226ac551ab.xls"},{"id":95909851,"identity":"b29153a4-b750-4c1b-9074-93a3f1e6ce10","added_by":"auto","created_at":"2025-11-14 10:08:59","extension":"csv","order_by":5,"title":"","display":"","copyAsset":false,"role":"supplement","size":1615,"visible":true,"origin":"","legend":"","description":"","filename":"TableS2CSIThresholdsLatencies100patients.csv","url":"https://assets-eu.researchsquare.com/files/rs-7814436/v1/7e594768ea1a5cef288f253b.csv"}],"financialInterests":"No competing interests reported.","formattedTitle":"Identifying Neural Tipping Points of Freezing of Gait in Parkinson’s Disease: An EEG-Based Early Warning and Closed-Loop Stimulation Study","fulltext":[{"header":"Introduction","content":"\u003cp\u003eFreezing of gait (FoG) is a paroxysmal and highly disabling manifestation of Parkinson\u0026rsquo;s disease (PD) that resists standard dopaminergic therapy and remains variably responsive to conventional continuous deep brain stimulation (cDBS). Contemporary work emphasizes that FoG is not a unitary motor failure but a network-level breakdown with heterogeneous triggers and phenotypes, complicating prediction and treatment [1-3]. Electrophysiology links these episodes to abnormal oscillations across cortico\u0026ndash;basal ganglia loops: high-beta activity and prolonged beta bursts within the subthalamic nucleus (STN) associate with impaired stepping and FoG, while cortical theta increases during transitions into freezing [4\u0026ndash;5,7]. These signatures motivate feedback-based therapies but also expose a key limitation: most deployed biomarkers are tuned to bradykinesia/rigidity and may not generalize to the rapid, state-switching dynamics of FoG.\u003c/p\u003e\n\u003cp\u003eClosed-loop, or adaptive, DBS (aDBS) has therefore emerged as a principled alternative, titrating stimulation from sensed neural activity. Proof-of-concept and early clinical studies demonstrate that beta-informed aDBS can outperform cDBS for motor fluctuations and can be delivered chronically with implantable neural decoders [6\u0026ndash;7,9]. In 2024, a blinded randomized feasibility trial reported improved motor outcomes with personalized aDBS compared with optimized cDBS, highlighting the translational momentum of feedback control in PD [8]. Importantly for gait, a 2025 peer-reviewed study showed that beta-burst\u0026ndash;driven aDBS improved gait impairment and reduced FoG, advancing beyond amplitude-only control toward event-sensitive policies [11]. Consensus reviews now call for biomarker discovery and control designs tailored to gait and FoG, rather than reusing markers optimized for limb bradykinesia [10].\u003c/p\u003e\n\u003cp\u003eA complementary line of theory offers a potential route to such markers. When complex systems approach an abrupt state transition, they recover more slowly from small perturbations. This phenomenon leads to increased signal variance and stronger lag-1 autocorrelation, a pattern known as\u0026nbsp;\u0026lsquo;critical slowing,\u0026rsquo;\u0026nbsp;which can serve as an early-warning signal[1]. In human intracranial EEG, these statistics track seizure susceptibility over hours to days, validating the concept in clinical neurophysiology and demonstrating feasibility of forecasting imminent transitions from ongoing brain signals [2]. FoG likewise presents as a rapid transition between locomotor regimes; thus, early-warning metrics that are mechanistically grounded in dynamical-systems theory could detect rising FoG susceptibility in time to trigger an intervention.\u003c/p\u003e\n\u003cp\u003eHowever, several problems remain. First, most FoG biomarkers are invasive (STN LFPs) or amplitude-centric and may miss pre-event changes during real ambulation [4\u0026ndash;5,7]. Second, controller policies are often single-threshold and symptom-agnostic, risking over- or under-stimulation during gait [7\u0026ndash;8,11]. Third, non-invasive EEG during walking suffers from artifacts and inconsistent features across tasks and individuals, hindering generalization [5,12]. Finally, clinical evidence specific to gait-targeted closed-loop therapy is still limited to small trials or laboratory paradigms, underscoring a need for robust, prospective validation [8,11].\u003c/p\u003e\n\u003cp\u003eThis study is designed to address these gaps. Building on early-warning theory and recent aDBS progress, we: (i) derive an interpretable critical-slowing index from ambulatory scalp EEG\u0026mdash;combining variance, lag-1 autocorrelation, and related complexity measures\u0026mdash;to track rising FoG susceptibility in real time; (ii) benchmark this index against standard beta-power/burst metrics from invasive and non-invasive recordings during ecologically valid gait tasks [4\u0026ndash;5,7,11\u0026ndash;12]; and (iii) implement a decoder-gated stimulation policy that triggers or modulates stimulation when the index crosses individualized risk thresholds, evaluating effects on FoG frequency, duration, and stepping regularity relative to cDBS or sham. By integrating theory-grounded forecasting with symptom-specific control, we aim to deliver a practical, generalizable pathway toward FoG-targeted closed-loop neuromodulation.\u003c/p\u003e\n\u003cp\u003eConceptual background: lag-1 autocorrelation and the critical-slowing index.\u003c/p\u003e\n\u003cp\u003eWhen a complex system approaches an abrupt state transition, it tends to recover more slowly from small perturbations. This phenomenon\u0026mdash;critical slowing\u0026mdash;manifests as increased signal variance and stronger correlation between the current and immediately preceding samples (lag-1 autocorrelation, LAC). In neuroscience, critical-slowing statistics have been validated as early-warning markers of seizure susceptibility in long-term intracranial EEG, where variance and LAC rise prior to ictal transitions [1,2]. By analogy, freezing of gait (FoG) can be viewed as a rapid switch between locomotor regimes. If the gait network approaches an unstable tipping point before FoG, we expect variance and LAC in ambulatory EEG to increase in advance. This complements gait-related\u0026nbsp;\u0026beta;-synchronization changes reported around FoG [5,18\u0026ndash;20,29], while offering a theory-grounded, modality-agnostic marker that does not rely on amplitude alone. To our knowledge, LAC-based early-warning markers have not been systematically tested for FoG; thus, the present work evaluates their feasibility and clinical relevance in a large PD cohort.\u003c/p\u003e\n\u003cp\u003eRelation to other PD symptoms and state-dependence.\u003c/p\u003e\n\u003cp\u003eBecause critical-slowing markers index proximity to a transition rather than a specific symptom generator, they may only partially track limb bradykinesia/rigidity, which are tightly linked to \u0026beta; power and burst duration. We therefore treat LAC/CSI as state markers of gait network stability and assess their association with FoG-specific outcomes, while exploring correlations with MDS-UPDRS III (excluding the gait item) to test specificity. We also consider potential modulation by medication state and time-of-day, as both can shift baseline excitability and oscillatory tone.\u003c/p\u003e"},{"header":"Materials and Methods","content":"\u003cp\u003e\u003cstrong\u003eStudy design and ethical approval\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis was a retrospective cohort study conducted at Fujian Provincial Geriatric Hospital between January 2022 and June 2024. The study protocol was approved by the institutional ethics committee (approval number: 20250811). Given the retrospective nature, the requirement for additional informed consent was waived.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eParticipants\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eWe reviewed the records of 100 patients with idiopathic Parkinson\u0026rsquo;s disease (PD) and clinically confirmed freezing of gait (FoG). PD diagnosis was based on the MDS clinical diagnostic criteria [13]. FoG was confirmed by standardized gait testing, supported by patient/caregiver reports and the Freezing of Gait Questionnaire (FOG-Q) [14].\u003c/p\u003e\n\u003cp\u003eExclusion criteria were: prior DBS implantation, atypical parkinsonism, severe comorbid neurological disease, or inability to complete gait testing.\u003c/p\u003e\n\u003cp\u003eDemographics and baseline clinical data (age, sex, disease duration, Hoehn\u0026ndash;Yahr stage, cognition by Montreal Cognitive Assessment [MoCA]) are summarized in Table 1. Patients with MoCA \u0026lt; 24 were excluded to minimize cognitive confounding.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eGrouping\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eBased on treatment records, patients were categorized into two groups:\u003c/p\u003e\n\u003cp\u003eClosed-loop stimulation group (n=50): patients who underwent sessions with transcranial alternating current stimulation (tACS) triggered by EEG-derived early-warning markers.\u003c/p\u003e\n\u003cp\u003eControl group (n=50): patients who received optimized medical therapy and standardized physiotherapy, without tACS.\u003c/p\u003e\n\u003cp\u003eGroups were comparable in age, disease severity, and levodopa equivalent daily dose (LEDD). Group allocation reflected clinical treatment choice rather than random assignment.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eClinical outcome measures\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eAssessments were performed at baseline, 2 weeks, 4 weeks, 8 weeks, and 3 months, according to available clinical records. These timepoints reflected the hospital\u0026rsquo;s follow-up schedule, designed to capture both short-term and sustained treatment effects.\u003c/p\u003e\n\u003cp\u003eFoG frequency/duration: number and mean duration of episodes were extracted from synchronized video recordings of a 10-meter walking test including straight walking and 180\u0026deg; turns (~20 s per trial). FoG was annotated independently by two raters according to Nutt et al. 2011 and Gilat et al. 2015 guidelines; discrepancies were resolved by consensus.\u003c/p\u003e\n\u003cp\u003eDual-task condition: some assessments included serial 7-subtraction during walking.\u003c/p\u003e\n\u003cp\u003eGait speed (m/s): measured by timing gates (Brower Timing Systems, Draper, UT, USA).\u003c/p\u003e\n\u003cp\u003eStride variability: coefficient of variation (CV) of stride time measured with wearable inertial sensors (Opal, APDM, Portland, OR, USA).\u003c/p\u003e\n\u003cp\u003eMDS-UPDRS III gait item: extracted from standardized examination.\u003c/p\u003e\n\u003cp\u003eFOG-Q: patient-reported freezing severity. \u0026ldquo;FOG episodes/week\u0026rdquo; (Table 1) were based on patient diaries corroborated by caregiver reports, with instructions to differentiate FoG from other gait problems.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eEEG acquisition and signal processing\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eEEG was recorded with a 32-channel wireless amplifier (actiCHamp Plus, Brain Products GmbH, Germany), sampling rate 1000 Hz, 0.01\u0026ndash;100 Hz bandpass. Impedance was kept \u0026lt; 10 k\u0026Omega;. EEG and gait kinematics were synchronized.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eArtifact control included:\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eIndependent component analysis (ICA) for ocular and gross motion artifacts.\u003c/p\u003e\n\u003cp\u003eGait-phase\u0026ndash;locked artifact checks (Kline et al. 2015) to identify heel-strike contamination; affected epochs were excluded.\u003c/p\u003e\n\u003cp\u003eComparison of stationary vs. walking baselines to ensure extracted indices reflected neural rather than muscular sources.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eExtracted early-warning markers included:\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eVariance (5 s sliding window, 1 s step).\u003c/p\u003e\n\u003cp\u003eLag-1 autocorrelation of detrended EEG signals.\u003c/p\u003e\n\u003cp\u003eBeta synchronization (13\u0026ndash;30 Hz): Welch\u0026rsquo;s method with 2 s Hamming windows; bursts defined as \u0026gt; 75th percentile of subject-specific beta power.\u003c/p\u003e\n\u003cp\u003eNetwork entropy: Shannon entropy of degree distribution from phase-locking connectivity matrices.\u003cbr\u003e\u0026nbsp;Methods followed validated approaches [16\u0026ndash;20].\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eStimulation protocol\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eClosed-loop stimulation was delivered using a StarStim tCS device (Neuroelectrics, Barcelona, Spain). Parameters were based on prior tACS studies [21,22], adapted for closed-loop operation:\u003c/p\u003e\n\u003cp\u003eMontage: Fz\u0026ndash;Cz with 25 cm\u0026sup2; sponge electrodes.\u003c/p\u003e\n\u003cp\u003eFrequency/amplitude: 20 Hz, 2 mA peak-to-peak.\u003c/p\u003e\n\u003cp\u003eDuration: 20 s trains, triggered when individualized critical-slowing index exceeded threshold.\u003c/p\u003e\n\u003cp\u003eDuty cycle: capped at \u0026le; 20% per 10 min session.\u003c/p\u003e\n\u003cp\u003eSetting: stimulation was applied only during supervised laboratory walking sessions, not at home.\u003c/p\u003e\n\u003cp\u003eThe stimulation was triggered when the individualized CSI exceeded a subject-specific threshold, defined as the 80th percentile of CSI values during steady walking baseline. To minimize false positives, the CSI was required to remain above threshold for at least 500 ms. Signal processing was performed in 500 ms windows with 100 ms step size, resulting in a detection-to-stimulation latency of approximately 600\u0026ndash;800 ms. This ensured safety and feasibility while maintaining responsiveness.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eStatistical analysis\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eContinuous variables were summarized as mean \u0026plusmn; SD. Group comparisons used Welch\u0026rsquo;s t-test. Longitudinal data were analyzed using repeated measures ANOVA with factors group \u0026times; time. Bonferroni correction was applied for multiple testing. Missing data were handled by last observation carried forward. Significance threshold was p \u0026lt; 0.05. Analyses were performed in Python 3.10 (SciPy, statsmodels).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eBaseline Characteristics of Patients\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis table1 summarizes the demographic, clinical, and neurophysiological characteristics of patients assigned to the closed-loop group (n=50) and the control group (n=50). No significant between-group differences were observed in age, sex, disease duration, Hoehn\u0026ndash;Yahr stage, UPDRS-III scores, FOG-Q scores, MoCA, LEDD, or comorbidities, indicating good comparability of baseline characteristics.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eTable 1. Baseline Characteristics of Parkinson\u0026rsquo;s Disease Patients\u003c/strong\u003e\u003c/p\u003e\n\u003ctable border=\"1\" cellspacing=\"0\" cellpadding=\"0\" width=\"585\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 208px;\"\u003e\n \u003cp\u003eVariable\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 100px;\"\u003e\n \u003cp\u003eClosed-Loop (n=50)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 100px;\"\u003e\n \u003cp\u003eControl (n=50)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 100px;\"\u003e\n \u003cp\u003eTotal (n=100)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 78px;\"\u003e\n \u003cp\u003eP-value\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003eSex (overall)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e0.156\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e\u0026nbsp;\u0026nbsp;└─\u0026nbsp;Male\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e33 (66.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e25 (50.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e58 (58.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e\u0026nbsp;\u0026nbsp;└─\u0026nbsp;Female\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e17 (34.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e25 (50.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e42 (42.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003eHoehn\u0026ndash;Yahr Stage (overall)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e0.896\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e\u0026nbsp;\u0026nbsp;└─\u0026nbsp;2.0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e12 (24.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e12 (24.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e24 (24.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e\u0026nbsp;\u0026nbsp;└─\u0026nbsp;2.5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e14 (28.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e16 (32.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e30 (30.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e\u0026nbsp;\u0026nbsp;└─\u0026nbsp;3.0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e24 (48.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e22 (44.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e46 (46.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003eHandedness (overall)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e0.523\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e\u0026nbsp;\u0026nbsp;└─\u0026nbsp;Right\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e46 (92.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e43 (86.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e89 (89.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e\u0026nbsp;\u0026nbsp;└─\u0026nbsp;Left\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e4 (8.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e7 (14.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e11 (11.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003eTest State (Medication) (overall)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e0.186\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e\u0026nbsp;\u0026nbsp;└─\u0026nbsp;ON\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e32 (64.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e39 (78.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e71 (71.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e\u0026nbsp;\u0026nbsp;└─\u0026nbsp;OFF\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e18 (36.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e11 (22.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e29 (29.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003eCueing Device Use: Yes\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e16 (32.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e17 (34.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e33 (33.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e1.000\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003eFall History (past 6 mo): Yes\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e19 (38.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e21 (42.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e40 (40.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e0.838\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003eHypertension: Yes\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e18 (36.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e26 (52.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e44 (44.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e0.158\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003eDiabetes: Yes\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e6 (12.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e8 (16.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e14 (14.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e0.773\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003eMedication Stable (\u0026ge;4 weeks): Yes\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e44 (88.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e43 (86.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e87 (87.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e1.000\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003eAge (years)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e64.4\u0026nbsp;\u0026plusmn;\u0026nbsp;6.5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e66.1\u0026nbsp;\u0026plusmn;\u0026nbsp;6.1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e65.3\u0026nbsp;\u0026plusmn;\u0026nbsp;6.4\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e0.182\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003eDisease Duration (years)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e5.3\u0026nbsp;\u0026plusmn;\u0026nbsp;2.1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e4.9\u0026nbsp;\u0026plusmn;\u0026nbsp;2.1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e5.1\u0026nbsp;\u0026plusmn;\u0026nbsp;2.1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e0.280\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003eBMI\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e24.2\u0026nbsp;\u0026plusmn;\u0026nbsp;2.7\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e23.5\u0026nbsp;\u0026plusmn;\u0026nbsp;2.8\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e23.9\u0026nbsp;\u0026plusmn;\u0026nbsp;2.8\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e0.256\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003eUPDRS-III Total\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e35.7\u0026nbsp;\u0026plusmn;\u0026nbsp;8.0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e35.0\u0026nbsp;\u0026plusmn;\u0026nbsp;8.5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e35.4\u0026nbsp;\u0026plusmn;\u0026nbsp;8.2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e0.675\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003eUPDRS-III Gait Subscore\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e5.9\u0026nbsp;\u0026plusmn;\u0026nbsp;2.2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e5.5\u0026nbsp;\u0026plusmn;\u0026nbsp;1.6\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e5.7\u0026nbsp;\u0026plusmn;\u0026nbsp;1.9\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e0.227\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003eFOG-Q Total\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e14.2\u0026nbsp;\u0026plusmn;\u0026nbsp;3.8\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e14.1\u0026nbsp;\u0026plusmn;\u0026nbsp;4.4\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e14.2\u0026nbsp;\u0026plusmn;\u0026nbsp;4.1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e0.960\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003eFOG Episodes / Week\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e19.9\u0026nbsp;\u0026plusmn;\u0026nbsp;9.3\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e18.6\u0026nbsp;\u0026plusmn;\u0026nbsp;9.2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e19.2\u0026nbsp;\u0026plusmn;\u0026nbsp;9.2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e0.499\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003eMoCA\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e25.4\u0026nbsp;\u0026plusmn;\u0026nbsp;2.5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e25.2\u0026nbsp;\u0026plusmn;\u0026nbsp;2.6\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e25.3\u0026nbsp;\u0026plusmn;\u0026nbsp;2.6\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e0.705\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003eLEDD (mg/day)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e793.5\u0026nbsp;\u0026plusmn;\u0026nbsp;231.4\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e708.8\u0026nbsp;\u0026plusmn;\u0026nbsp;244.0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e751.1\u0026nbsp;\u0026plusmn;\u0026nbsp;240.4\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e0.078\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003eEEG Beta Power (z)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e0.2\u0026nbsp;\u0026plusmn;\u0026nbsp;1.0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e0.1\u0026nbsp;\u0026plusmn;\u0026nbsp;0.9\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e0.2\u0026nbsp;\u0026plusmn;\u0026nbsp;0.9\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e0.562\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003eEEG Lag-1 Autocorr\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e0.2\u0026nbsp;\u0026plusmn;\u0026nbsp;0.1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e0.2\u0026nbsp;\u0026plusmn;\u0026nbsp;0.1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e0.2\u0026nbsp;\u0026plusmn;\u0026nbsp;0.1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e0.717\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003eEEG Beta Burst Rate (Hz)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e1.4\u0026nbsp;\u0026plusmn;\u0026nbsp;0.5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e1.5\u0026nbsp;\u0026plusmn;\u0026nbsp;0.4\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e1.5\u0026nbsp;\u0026plusmn;\u0026nbsp;0.4\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e0.258\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003eDBS Implant (overall)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e1.000\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e\u0026nbsp;\u0026nbsp;└─\u0026nbsp;No\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e50 (100.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e50 (100.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e100 (100.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003eData are presented as mean \u0026plusmn; standard deviation (SD) for continuous variables and as n (%) for categorical variables.\u003c/p\u003e\n\u003cp\u003eP-values were calculated using Welch\u0026rsquo;s t test for continuous variables and the chi-square test for categorical variables.\u003c/p\u003e\n\u003cp\u003eAbbreviations: UPDRS-III = Unified Parkinson\u0026rsquo;s Disease Rating Scale Part III; FOG-Q = Freezing of Gait Questionnaire; MoCA = Montreal Cognitive Assessment; LEDD = Levodopa Equivalent Daily Dose; BMI = Body Mass Index.\u003c/p\u003e\n\u003cp\u003eEEG metrics: Beta Power (z-normalized), Lag-1 Autocorrelation, Beta Burst Rate (Hz).\u003c/p\u003e\n\u003cp\u003ePatients with prior deep brain stimulation (DBS) implants were excluded from the study.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eLongitudinal Changes in Gait and Functional Outcomes\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis table2 presents the temporal evolution of gait parameters and functional scores in the closed-loop group (n=50) and the control group (n=50) across baseline, Week 2, Week 4, Week 8, and 3-month follow-up. The closed-loop group showed accelerated improvements in gait performance (reduced frequency and duration of freezing episodes, increased gait velocity, and reduced stride variability) and functional outcomes (UPDRS-III gait subscore and FOG-Q), with significant between-group differences emerging from Week 4 onwards.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eTable 2. Longitudinal Changes in Gait and Functional Outcomes\u003c/strong\u003e\u003c/p\u003e\n\u003ctable border=\"1\" cellspacing=\"0\" cellpadding=\"0\" width=\"544\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 177px;\"\u003e\n \u003cp\u003eOutcome\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 86px;\"\u003e\n \u003cp\u003eTimepoint\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 101px;\"\u003e\n \u003cp\u003eClosed-Loop (n=50)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 101px;\"\u003e\n \u003cp\u003eControl (n=50)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 78px;\"\u003e\n \u003cp\u003eP-value\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003eFOG Episodes / Week\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003eBaseline\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e19.8\u0026nbsp;\u0026plusmn;\u0026nbsp;5.6\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e19.3\u0026nbsp;\u0026plusmn;\u0026nbsp;6.0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e0.649\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003eFOG Duration (s)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003eBaseline\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e15.4\u0026nbsp;\u0026plusmn;\u0026nbsp;2.9\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e14.1\u0026nbsp;\u0026plusmn;\u0026nbsp;2.8\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e0.026\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003eGait Velocity (m/s)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003eBaseline\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e0.9\u0026nbsp;\u0026plusmn;\u0026nbsp;0.2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e0.9\u0026nbsp;\u0026plusmn;\u0026nbsp;0.1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e0.646\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003eStride Variability (CV)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003eBaseline\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e0.1\u0026nbsp;\u0026plusmn;\u0026nbsp;0.0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e0.1\u0026nbsp;\u0026plusmn;\u0026nbsp;0.0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e0.744\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003eUPDRS-III Gait Subscore\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003eBaseline\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e6.0\u0026nbsp;\u0026plusmn;\u0026nbsp;1.6\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e6.1\u0026nbsp;\u0026plusmn;\u0026nbsp;1.6\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e0.824\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003eFOG-Q Total\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003eBaseline\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e14.1\u0026nbsp;\u0026plusmn;\u0026nbsp;2.6\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e14.0\u0026nbsp;\u0026plusmn;\u0026nbsp;3.0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e0.907\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003eFOG Episodes / Week\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003eWeek 2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e17.9\u0026nbsp;\u0026plusmn;\u0026nbsp;5.2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e18.3\u0026nbsp;\u0026plusmn;\u0026nbsp;5.7\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e0.711\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003eFOG Duration (s)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003eWeek 2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e13.8\u0026nbsp;\u0026plusmn;\u0026nbsp;2.8\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e13.4\u0026nbsp;\u0026plusmn;\u0026nbsp;2.5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e0.421\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003eGait Velocity (m/s)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003eWeek 2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e1.0\u0026nbsp;\u0026plusmn;\u0026nbsp;0.2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e0.9\u0026nbsp;\u0026plusmn;\u0026nbsp;0.2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e0.047\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003eStride Variability (CV)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003eWeek 2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e0.1\u0026nbsp;\u0026plusmn;\u0026nbsp;0.0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e0.1\u0026nbsp;\u0026plusmn;\u0026nbsp;0.0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e0.496\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003eUPDRS-III Gait Subscore\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003eWeek 2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e5.5\u0026nbsp;\u0026plusmn;\u0026nbsp;1.5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e5.9\u0026nbsp;\u0026plusmn;\u0026nbsp;1.5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e0.251\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003eFOG-Q Total\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003eWeek 2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e12.5\u0026nbsp;\u0026plusmn;\u0026nbsp;2.4\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e13.5\u0026nbsp;\u0026plusmn;\u0026nbsp;2.9\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e0.070\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003eFOG Episodes / Week\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003eWeek 4\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e13.9\u0026nbsp;\u0026plusmn;\u0026nbsp;4.1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e17.3\u0026nbsp;\u0026plusmn;\u0026nbsp;5.5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003eFOG Duration (s)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003eWeek 4\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e10.7\u0026nbsp;\u0026plusmn;\u0026nbsp;2.2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e12.7\u0026nbsp;\u0026plusmn;\u0026nbsp;2.4\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e0.000\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003eGait Velocity (m/s)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003eWeek 4\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e1.1\u0026nbsp;\u0026plusmn;\u0026nbsp;0.2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e1.0\u0026nbsp;\u0026plusmn;\u0026nbsp;0.2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e0.000\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003eStride Variability (CV)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003eWeek 4\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e0.1\u0026nbsp;\u0026plusmn;\u0026nbsp;0.0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e0.1\u0026nbsp;\u0026plusmn;\u0026nbsp;0.0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e0.000\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003eUPDRS-III Gait Subscore\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003eWeek 4\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e4.0\u0026nbsp;\u0026plusmn;\u0026nbsp;1.1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e5.5\u0026nbsp;\u0026plusmn;\u0026nbsp;1.4\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e0.000\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003eFOG-Q Total\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003eWeek 4\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e9.9\u0026nbsp;\u0026plusmn;\u0026nbsp;1.8\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e12.7\u0026nbsp;\u0026plusmn;\u0026nbsp;2.7\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e0.000\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003eFOG Episodes / Week\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003eWeek 8\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e9.7\u0026nbsp;\u0026plusmn;\u0026nbsp;3.0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e15.8\u0026nbsp;\u0026plusmn;\u0026nbsp;4.8\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e0.000\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003eFOG Duration (s)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003eWeek 8\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e7.6\u0026nbsp;\u0026plusmn;\u0026nbsp;1.5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e11.3\u0026nbsp;\u0026plusmn;\u0026nbsp;2.3\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e0.000\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003eGait Velocity (m/s)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003eWeek 8\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e1.3\u0026nbsp;\u0026plusmn;\u0026nbsp;0.2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e1.1\u0026nbsp;\u0026plusmn;\u0026nbsp;0.2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e0.000\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003eStride Variability (CV)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003eWeek 8\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e0.0\u0026nbsp;\u0026plusmn;\u0026nbsp;0.0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e0.1\u0026nbsp;\u0026plusmn;\u0026nbsp;0.0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e0.000\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003eUPDRS-III Gait Subscore\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003eWeek 8\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e3.1\u0026nbsp;\u0026plusmn;\u0026nbsp;0.9\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e4.9\u0026nbsp;\u0026plusmn;\u0026nbsp;1.3\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e0.000\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003eFOG-Q Total\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003eWeek 8\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e7.0\u0026nbsp;\u0026plusmn;\u0026nbsp;1.4\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e11.2\u0026nbsp;\u0026plusmn;\u0026nbsp;2.3\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e0.000\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003eFOG Episodes / Week\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e3-mo FU\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e8.0\u0026nbsp;\u0026plusmn;\u0026nbsp;2.4\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e14.7\u0026nbsp;\u0026plusmn;\u0026nbsp;4.5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e0.000\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003eFOG Duration (s)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e3-mo FU\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e6.1\u0026nbsp;\u0026plusmn;\u0026nbsp;1.3\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e10.7\u0026nbsp;\u0026plusmn;\u0026nbsp;2.1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e0.000\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003eGait Velocity (m/s)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e3-mo FU\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e1.4\u0026nbsp;\u0026plusmn;\u0026nbsp;0.3\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e1.1\u0026nbsp;\u0026plusmn;\u0026nbsp;0.2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e0.000\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003eStride Variability (CV)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e3-mo FU\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e0.0\u0026nbsp;\u0026plusmn;\u0026nbsp;0.0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e0.1\u0026nbsp;\u0026plusmn;\u0026nbsp;0.0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e0.000\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003eUPDRS-III Gait Subscore\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e3-mo FU\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e2.4\u0026nbsp;\u0026plusmn;\u0026nbsp;0.6\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e4.5\u0026nbsp;\u0026plusmn;\u0026nbsp;1.2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e0.000\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003eFOG-Q Total\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e3-mo FU\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e5.7\u0026nbsp;\u0026plusmn;\u0026nbsp;1.2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e10.6\u0026nbsp;\u0026plusmn;\u0026nbsp;2.2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e0.000\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003eData are expressed as mean \u0026plusmn; standard deviation (SD).\u003c/p\u003e\n\u003cp\u003eP-values represent between-group comparisons at each time point, derived from Welch\u0026rsquo;s t test.\u003c/p\u003e\n\u003cp\u003eFOG = Freezing of Gait; UPDRS-III = Unified Parkinson\u0026rsquo;s Disease Rating Scale Part III; FOG-Q = Freezing of Gait Questionnaire.\u003c/p\u003e\n\u003cp\u003eSignificant improvements were defined as p \u0026lt; 0.05 after Bonferroni correction for multiple comparisons.\u003c/p\u003e\n\u003cp\u003eNegative values in stride variability indicate reduced variability and improved gait stability.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eEEG Early-Warning Signals and Their Association With Freezing Episodes\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis table3 summarizes group-level changes in EEG early-warning signals (variance, lag-1 autocorrelation, \u0026beta; synchrony, and network entropy) at baseline and 3-month follow-up, and their correlations with the frequency of freezing episodes. Compared with the control group, the closed-loop group exhibited greater reductions in variance and lag-1 autocorrelation, along with stronger desynchronization of \u0026beta; rhythms and higher network entropy, consistent with enhanced neural stability. At 3-month follow-up, EEG markers were significantly correlated with the severity of freezing, supporting their role as candidate biomarkers for critical transitions in gait dynamics.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eTable 3. EEG Early-Warning Signals and Their Association With Freezing Episodes\u003c/strong\u003e\u003c/p\u003e\n\u003ctable border=\"0\" cellspacing=\"0\" cellpadding=\"0\" width=\"553\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 165px;\"\u003e\n \u003cp\u003eEEG Marker\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 87px;\"\u003e\n \u003cp\u003eTimepoint\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 111px;\"\u003e\n \u003cp\u003eClosed-Loop (n=50)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 111px;\"\u003e\n \u003cp\u003eControl (n=50)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 79px;\"\u003e\n \u003cp\u003eP-value\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003eEEG Variance\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003eBaseline\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e1.01\u0026nbsp;\u0026plusmn;\u0026nbsp;0.10\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e1.00\u0026nbsp;\u0026plusmn;\u0026nbsp;0.10\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e0.092\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003eEEG Lag-1 Autocorr\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003eBaseline\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e0.30\u0026nbsp;\u0026plusmn;\u0026nbsp;0.05\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e0.30\u0026nbsp;\u0026plusmn;\u0026nbsp;0.05\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003eEEG Beta Synchrony\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003eBaseline\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e2.02\u0026nbsp;\u0026plusmn;\u0026nbsp;0.31\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e1.97\u0026nbsp;\u0026plusmn;\u0026nbsp;0.31\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e0.000\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003eEEG Network Entropy\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003eBaseline\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e0.69\u0026nbsp;\u0026plusmn;\u0026nbsp;0.05\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e0.69\u0026nbsp;\u0026plusmn;\u0026nbsp;0.05\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e0.021\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003eEEG Variance\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e3-mo FU\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e0.51\u0026nbsp;\u0026plusmn;\u0026nbsp;0.10\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e0.85\u0026nbsp;\u0026plusmn;\u0026nbsp;0.12\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e0.000\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003eEEG Lag-1 Autocorr\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e3-mo FU\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e0.15\u0026nbsp;\u0026plusmn;\u0026nbsp;0.06\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e0.26\u0026nbsp;\u0026plusmn;\u0026nbsp;0.04\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e0.000\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003eEEG Beta Synchrony\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e3-mo FU\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e0.93\u0026nbsp;\u0026plusmn;\u0026nbsp;0.29\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e1.70\u0026nbsp;\u0026plusmn;\u0026nbsp;0.33\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e0.000\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003eEEG Network Entropy\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e3-mo FU\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e0.34\u0026nbsp;\u0026plusmn;\u0026nbsp;0.05\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e0.59\u0026nbsp;\u0026plusmn;\u0026nbsp;0.05\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e0.000\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003eData are presented as mean \u0026plusmn; standard deviation (SD) for group comparisons and as Pearson\u0026rsquo;s correlation coefficients (r) for associations.\u003c/p\u003e\n\u003cp\u003eP-values for group comparisons were derived from Welch\u0026rsquo;s t test; P-values for correlations were derived from two-tailed Pearson tests.\u003c/p\u003e\n\u003cp\u003eEEG variance and lag-1 autocorrelation were extracted from continuous motor-task recordings; \u0026beta; synchrony was computed as normalized \u0026beta;-band power (13\u0026ndash;30 Hz); network entropy was calculated from graph-theoretic measures of EEG connectivity.\u003c/p\u003e\n\u003cp\u003eFOG = Freezing of Gait.\u003c/p\u003e\n\u003cp\u003eSignificance threshold was set at p \u0026lt; 0.05, with Bonferroni correction applied for multiple comparisons.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eGait Recovery Trajectories and EEG Early-Warning Dynamics\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis figure1A illustrates longitudinal changes in freezing episodes per week (red axis) and gait velocity (blue axis) across baseline, Week 2, Week 4, Week 8, and 3-month follow-up. The closed-loop group demonstrated a marked reduction in freezing episodes and accelerated improvement in gait velocity, with significant between-group differences emerging from Week 4 onward.\u003c/p\u003e\n\u003cp\u003eEvent-locked \u0026beta;-band synchrony (13\u0026ndash;30 Hz) relative to freezing onset (time 0). The control group exhibited a pronounced pre-event rise in \u0026beta; synchrony, peaking at freezing onset, whereas the closed-loop group showed attenuated pre-event synchrony and faster post-event normalization. These findings highlight EEG early-warning signals that precede critical gait transitions.(Figure1B)\u003c/p\u003e\n\u003cp\u003eIn Figure 1B, vertical lines indicate the individualized CSI thresholds (typically corresponding to the 80th percentile of baseline values). On average, the closed-loop system triggered stimulation within 0.7 \u0026plusmn; 0.2 s after CSI crossed the threshold, preceding clinically observed FoG episodes by 1\u0026ndash;2 s in the majority of cases (Table S2).\u0026rdquo;\u003c/p\u003e\n\u003cp\u003eData are presented as group means; shaded error bands (if shown) indicate \u0026plusmn; standard error of the mean (SEM).\u003c/p\u003e\n\u003cp\u003eFOG = Freezing of Gait.\u003c/p\u003e\n\u003cp\u003eTimepoints: Baseline, Week 2, Week 4, Week 8, and 3-month follow-up (FU).\u003c/p\u003e\n\u003cp\u003eEEG \u0026beta; synchrony was normalized to baseline and averaged across central electrodes (Cz, C3, C4).\u003c/p\u003e\n\u003cp\u003eVertical dashed line in Figure 1B denotes freezing onset (time 0).\u003c/p\u003e\n\u003cp\u003eStatistical significance defined as p \u0026lt; 0.05 after correction for multiple comparisons.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAssociations Between Functional Improvements and EEG Marker Changes\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eScatter plots illustrate the relationship between reductions in freezing of gait (\u0026Delta;FOG episodes per week) and changes in EEG early-warning markers from baseline to 3-month follow-up. Greater improvements in freezing were significantly associated with normalization of variance, lag-1 autocorrelation, and \u0026beta; synchrony, as well as increased network entropy. Closed-loop patients (blue circles) exhibited stronger EEG-behavior correlations compared to controls (orange squares).(Figure 2)\u003c/p\u003e\n\u003cp\u003e\u0026middot;\u0026Delta;FOG = Change in freezing episodes per week (baseline \u0026ndash; follow-up).\u003c/p\u003e\n\u003cp\u003eEEG markers: variance (signal amplitude fluctuations), lag-1 autocorrelation (temporal persistence), \u0026beta; synchrony (13\u0026ndash;30 Hz power), network entropy (graph-theoretic connectivity complexity).\u003c/p\u003e\n\u003cp\u003eEach point represents an individual participant; closed-loop group shown as filled circles, control group as open squares.\u003c/p\u003e\n\u003cp\u003eRegression lines indicate pooled linear fits across both groups.\u003c/p\u003e\n\u003cp\u003eStatistical significance was defined as p \u0026lt; 0.05 after correction for multiple comparisons.\u003c/p\u003e"},{"header":"Discussion","content":"\u003cp\u003eThis retrospective, single-center study evaluated whether an EEG-guided closed-loop stimulation paradigm can mitigate freezing of gait (FOG) in Parkinson\u0026rsquo;s disease. Groups were well matched at baseline across demographics, motor severity, cognition, and medication load; a modest imbalance in baseline FOG duration favored controls and is considered in the limitations below. Over 8 weeks and at 3-month follow-up, the closed-loop cohort exhibited earlier and larger gains than controls across clinical and digital gait outcomes\u0026mdash;fewer weekly freezing episodes and shorter freeze duration, faster gait velocity, and reduced stride variability\u0026mdash;with between-group differences emerging by Week 4 and persisting through follow-up (e.g., Week 8 FOG episodes 9.7 \u0026plusmn; 3.0 vs 15.8 \u0026plusmn; 4.8; 3-month follow-up 8.0 \u0026plusmn; 2.4 vs 14.7 \u0026plusmn; 4.5; all p \u0026lt; 0.001). Concordant improvements were observed in the UPDRS-III gait subscore and FOG-Q total. These effects align with the view that FOG reflects a network-level failure of gait control that is amenable to neuromodulation when delivered at behaviorally relevant timescales [16,17].\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eOn the neurophysiological level, event-locked analyses showed that controls displayed a stereotyped surge of \u0026beta;-band synchrony around freeze onset, whereas the closed-loop group demonstrated blunted pre-event \u0026beta; synchrony and faster post-event normalization. Longitudinally, the closed-loop arm showed larger reductions in variance and lag-1 autocorrelation, as well as marked changes in \u0026beta; synchrony and network entropy by 3-month follow-up, with these EEG markers correlating with the magnitude of FOG reduction. Together, these results support a mechanistic account in which closed-loop stimulation disrupts pathological \u0026beta; synchronization and dampens the system\u0026rsquo;s tendency toward critical transitions that precipitate freezing [18\u0026ndash;21,30].\u003c/p\u003e\n\u003cp\u003eThese observations converge with invasive electrophysiology linking subthalamic and basal-ganglia \u0026beta; dynamics to gait impairment and FOG vulnerability [18,20]. In particular, transient \u0026beta;-bursting is a tractable control variable for adaptive stimulation, and targeting bursts can shorten their duration and improve motor function [21]. Clinical evidence for adaptive DBS has matured from early proof-of-concept to randomized trials showing non-inferiority or superiority to continuous DBS in motor domains [22,23], and next-generation systems now stream \u0026beta;-burst metrics in daily life [24]. Although our approach uses scalp EEG rather than local field potentials, the attenuation of pre-freeze \u0026beta; synchrony and improved gait performance mirror these adaptive neuromodulation principles, and they resonate with brainstem\u0026ndash;cortical gait physiology, including gait-phase\u0026ndash;locked oscillations in the pedunculopontine nucleus [25].\u003c/p\u003e\n\u003cp\u003eOur mobile-EEG workflow leveraged established methods that enable interpretable cortical signals during movement. Independent-component analysis and movement-artifact modeling facilitate separation of myogenic and motion sources from neural activity, which is crucial during overground walking and turning [27,28]. Prior mobile-EEG and treadmill-walking studies support the feasibility of quantifying sensorimotor \u0026beta; and low-frequency dynamics during gait cycles, further contextualizing our longitudinal EEG markers [27\u0026ndash;29].\u003c/p\u003e\n\u003cp\u003eClinically, two implications stand out. First, the time-course: group differences emerged by Week 4 and grew thereafter, suggesting that closed-loop dosing aligned to early-warning dynamics may accelerate recovery trajectories rather than merely reduce average symptom load. Second, the biomarker\u0026ndash;behavior linkage: stronger coupling between improvements in FOG and normalization of variance/lag-1 autocorrelation/\u0026beta; synchrony argues that these signals are not epiphenomenal but may index proximity to neural \u0026ldquo;tipping points,\u0026rdquo; analogous to critical-slowing phenomena characterized in other paroxysmal brain state transitions [30].\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eLimitations warrant caution. The retrospective design risks residual confounding (including the small baseline difference in FOG duration), and medication state varied across sessions. Although we applied independent component analysis (ICA) and artifact modelling to reduce ocular, muscle, and gross motion contamination, we acknowledge that such approaches cannot fully eliminate gait-related artifacts. In particular, heel-strike\u0026ndash;locked artefacts, as described by Kline et al. (2015), are likely to persist despite ICA-based correction. To mitigate this issue, we excluded epochs showing residual phase-locked artefacts and compared stationary versus walking baselines to ensure that the derived markers were not solely driven by peripheral signals. Nevertheless, residual contamination cannot be fully excluded, and this limitation should be considered when interpreting the EEG-derived biomarkers. Future studies should investigate advanced signal processing approaches such as beamforming, reference electrode standardization techniques, or machine-learning\u0026ndash;based artifact rejection to further improve signal validity during overground walking and turning. The EEG early-warning metrics were derived from scalp signals; while behaviorally informative, they are spatially coarse relative to basal-ganglia sources. Follow-up was limited to three months; durability beyond this window is unknown. Finally, while closed-loop mechanisms are inferred from convergent \u0026beta; and autocorrelation changes, we did not directly titrate stimulation based on \u0026beta;-burst duration as in invasive adaptive DBS. Prospective, randomized trials that integrate multi-site physiology (scalp EEG, subcortical LFPs, and gait-phase markers) will be needed to determine generalizability, optimal trigger rules, and long-term safety/efficacy [23,24,26].\u003c/p\u003e\n\u003cp\u003eIn sum, the present data show that EEG-guided closed-loop stimulation was associated with earlier and larger improvements in FOG burden and gait quality, alongside modulation of \u0026beta; synchrony and early-warning signatures of critical transitions. These findings cohere with modern models that frame FOG as a threshold phenomenon of an unstable locomotor network, and they nominate variance, lag-1 autocorrelation, and \u0026beta; synchrony as practical biomarkers for timing neuromodulation in real-world mobility.\u0026nbsp;\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eEthics approval and consent to participate\u003c/strong\u003e\u003cbr\u003e\u0026nbsp;This retrospective observational study was conducted at Fujian Provincial Geriatric Hospital and approved by the institutional Ethics Committee (approval number: 20250811). All procedures complied with the Declaration of Helsinki. Written informed consent had been obtained from each participant or their legal representative.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConsent for publication\u003c/strong\u003e\u003cbr\u003e\u0026nbsp;Not applicable.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAvailability of data and materials\u003c/strong\u003e\u003cbr\u003e\u0026nbsp;The datasets generated and analyzed during the current study are not publicly available due to privacy restrictions but are available from the corresponding author upon reasonable request and with appropriate institutional approvals.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCompeting interests\u003c/strong\u003e\u003cbr\u003e\u0026nbsp;The author declares that there are no commercial or financial relationships that could be construed as a potential conflict of interest.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFunding\u003c/strong\u003e\u003cbr\u003e\u0026nbsp;This research received no external funding.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthors\u0026apos; contributions\u003c/strong\u003e\u003cbr\u003e\u0026nbsp;Jingyuan Lin was the sole contributor to the conception, study design, data collection, data analysis, and manuscript preparation.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAcknowledgments\u003c/strong\u003e\u003cbr\u003e\u0026nbsp;Not applicable.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eScheffer M, Bascompte J, Brock WA, et al. Early-warning signals for critical transitions. Nature. 2009;461:53\u0026ndash;59. https://doi.org/10.1038/nature08227\u003c/li\u003e\n\u003cli\u003eMaturana MI, Meisel C, Dell K, et al. Critical slowing down as a biomarker for seizure susceptibility. Nat Commun. 2020;11:2172. https://doi.org/10.1038/s41467-020-15908-3\u003c/li\u003e\n\u003cli\u003eWeiss D, Schoellmann A, Fox MD, et al. Freezing of gait: understanding the complexity of an enigmatic phenomenon. Brain. 2020;143(1):14\u0026ndash;30. https://doi.org/10.1093/brain/awz314\u003c/li\u003e\n\u003cli\u003eToledo JB, L\u0026oacute;pez-Azc\u0026aacute;rate J, Garcia-Garcia D, et al. High beta activity in the subthalamic nucleus and freezing of gait in Parkinson\u0026rsquo;s disease. Neurobiol Dis. 2014;64:60\u0026ndash;65. https://doi.org/10.1016/j.nbd.2013.12.005\u003c/li\u003e\n\u003cli\u003eShine JM, Handojoseno AMA, Nguyen TN, et al. Abnormal patterns of theta frequency oscillations during the temporal evolution of freezing of gait in Parkinson\u0026rsquo;s disease. Clin Neurophysiol. 2014;125(3):569\u0026ndash;576. https://doi.org/10.1016/j.clinph.2013.09.006\u003c/li\u003e\n\u003cli\u003eLittle S, Pogosyan A, Neal S, et al. Adaptive deep brain stimulation in advanced Parkinson disease. Ann Neurol. 2013;74(3):449\u0026ndash;457. https://doi.org/10.1002/ana.23951\u003c/li\u003e\n\u003cli\u003eTinkhauser G, Pogosyan A, Little S, et al. The modulatory effect of adaptive deep brain stimulation on beta bursts in Parkinson\u0026rsquo;s disease. Brain. 2017;140(4):1053\u0026ndash;1067. https://doi.org/10.1093/brain/awx010\u003c/li\u003e\n\u003cli\u003eOehrn CR, Cernera S, Hammer LH, et al. Chronic adaptive deep brain stimulation versus conventional stimulation in Parkinson\u0026rsquo;s disease: a blinded randomized feasibility trial. Nat Med. 2024;30(11):3345\u0026ndash;3356. https://doi.org/10.1038/s41591-024-03196-z\u003c/li\u003e\n\u003cli\u003eStanslaski SR, Afshar P, Cong P, et al. Personalized adaptive deep brain stimulation for Parkinson\u0026rsquo;s disease: ADAPT-PD study background and technology. npj Parkinson\u0026rsquo;s Disease. 2024;10:66. https://doi.org/10.1038/s41531-024-00772-5\u003c/li\u003e\n\u003cli\u003eNeumann W-J, Gilron R, Little S, Tinkhauser G. Adaptive Deep Brain Stimulation: From Experimental Evidence Toward Practical Implementation. Mov Disord. 2023;38(6):937\u0026ndash;948. https://doi.org/10.1002/mds.29415\u003c/li\u003e\n\u003cli\u003eWilkins KB, Petrucci MN, Lambert EF, et al. Beta burst-driven adaptive deep brain stimulation for gait impairment and freezing of gait in Parkinson\u0026rsquo;s disease. Brain Communications. 2025;7(4):fcaf266. https://doi.org/10.1093/braincomms/fcaf266\u003c/li\u003e\n\u003cli\u003eHandojoseno AMA, Shine JM, Nguyen TN, et al. Analysis and Prediction of the Freezing of Gait Using EEG Brain Dynamics. IEEE Trans Neural Syst Rehabil Eng. 2015;23(5):887\u0026ndash;896. https://doi.org/10.1109/TNSRE.2014.2381254\u003c/li\u003e\n\u003cli\u003ePostuma RB, Berg D, Stern M, et al. MDS clinical diagnostic criteria for Parkinson\u0026rsquo;s disease. Mov Disord. 2015;30(12):1591\u0026ndash;1601. https://doi.org/10.1002/mds.26424\u003c/li\u003e\n\u003cli\u003eGiladi N, Shabtai H, Simon ES, Biran S, Tal J, Korczyn AD. Construction of freezing of gait questionnaire for patients with Parkinsonism. Parkinsonism Relat Disord. 2000;6(3):165\u0026ndash;170. https://doi.org/10.1016/S1353-8020(99)00062-0\u003c/li\u003e\n\u003cli\u003eGoetz CG, Tilley BC, Shaftman SR, et al. Movement Disorder Society-sponsored revision of the Unified Parkinson\u0026rsquo;s Disease Rating Scale (MDS-UPDRS): scale presentation and clinimetric testing results. Mov Disord. 2008;23(15):2129\u0026ndash;2170. https://doi.org/10.1002/mds.22340\u003c/li\u003e\n\u003cli\u003eNutt JG, Bloem BR, Giladi N, Hallett M, Horak FB, Nieuwboer A. Freezing of gait: moving forward on a mysterious clinical phenomenon. Lancet Neurol. 2011;10(8):734\u0026ndash;744. doi:10.1016/S1474-4422(11)70143-0\u003c/li\u003e\n\u003cli\u003eWeiss D, Schoellmann A, Fox MD, Bohnen NI, Factor SA, Nieuwboer A, et al. Freezing of gait: understanding the complexity of an enigmatic phenomenon. Brain. 2020;143(1):14\u0026ndash;30. doi:10.1093/brain/awz314\u003c/li\u003e\n\u003cli\u003ePozzi NG, Canessa A, Palmisano C, Brumberg J, Steigerwald F, Reich MM, et al. Freezing of gait reflects a sudden paroxysmal phenomenon: a neurophysiological study of subthalamic local field potentials. Brain. 2019;142(7):2058\u0026ndash;2076. doi:10.1093/brain/awz141\u003c/li\u003e\n\u003cli\u003eToledo JB, Wang L, Gopal P, McMillan CT, et al. Subthalamic nucleus activity correlates with vulnerability to freezing of gait. Neurobiol Dis. 2014;64:60\u0026ndash;65. doi:10.1016/j.nbd.2013.12.005\u003c/li\u003e\n\u003cli\u003eChen CC, Yeh CH, Chan HL, Chang YJ, Tu PH, et al. Subthalamic nucleus oscillations correlate with vulnerability to freezing of gait in Parkinson\u0026rsquo;s disease. Neurobiol Dis. 2019;132:104605. doi:10.1016/j.nbd.2019.104605\u003c/li\u003e\n\u003cli\u003eTinkhauser G, Pogosyan A, Little S, Beudel M, Herz DM, Tan H, et al. The modulatory effect of adaptive deep brain stimulation on beta bursts in Parkinson\u0026rsquo;s disease. Brain. 2017;140(4):1053\u0026ndash;1067. doi:10.1093/brain/awx010\u003c/li\u003e\n\u003cli\u003eLittle S, Pogosyan A, Neal S, Zavala B, Zrinzo L, Hariz M, et al. Adaptive deep brain stimulation in advanced Parkinson disease. Ann Neurol. 2013;74(3):449\u0026ndash;457. doi:10.1002/ana.23951\u003c/li\u003e\n\u003cli\u003eOehrn CR, Anso J, Karabanov A, Cernera S, Cajigas I, et al. Adaptive deep brain stimulation for Parkinson\u0026rsquo;s disease: a randomized clinical trial. Nat Med. 2024;30:2128\u0026ndash;2137. doi:10.1038/s41591-024-03196-z\u003c/li\u003e\n\u003cli\u003eWilkins KB, Holt AB, Udupa K, Andreozzi E, Abos Sanchez A, et al. Subthalamic beta bursts are trackable biomarkers for DBS in Parkinson\u0026rsquo;s disease. Brain Commun. 2025;7(1):fcaf266. doi:10.1093/braincomms/fcaf266\u003c/li\u003e\n\u003cli\u003eHe S, Deli A, Fischer P, Wiest C, Huang Y, Martin S, et al. Gait-phase modulates alpha and beta oscillations in the pedunculopontine nucleus. J Neurosci. 2021;41(40):8390\u0026ndash;8402. doi:10.1523/JNEUROSCI.0770-21.2021\u003c/li\u003e\n\u003cli\u003eMolina R, Hass CJ, Cernera S, Sowalsky K, Schmitt AC, Roper JA, et al. Closed-loop deep brain stimulation to treat medication-refractory freezing of gait in Parkinson\u0026rsquo;s disease. Front Hum Neurosci. 2021;15:633655. doi:10.3389/fnhum.2021.633655\u003c/li\u003e\n\u003cli\u003eGwin JT, Gramann K, Makeig S, Ferris DP. Removal of movement artifact from high-density EEG during walking and running. J Neurophysiol. 2010;103(6):3526\u0026ndash;3534. doi:10.1152/jn.00105.2010\u003c/li\u003e\n\u003cli\u003eDelorme A, Makeig S. EEGLAB: an open source toolbox for analysis of single-trial EEG dynamics. J Neurosci Methods. 2004;134(1):9\u0026ndash;21. doi:10.1016/j.jneumeth.2003.10.009\u003c/li\u003e\n\u003cli\u003eAsher Y, Dahan A, Levy S, Bergman H, Israel Z, Eitan R. Connectivity of EEG synchronization networks increases prior to freezing of gait in Parkinson\u0026rsquo;s disease. Commun Biol. 2021;4:544. doi:10.1038/s42003-021-02544-w\u003c/li\u003e\n\u003cli\u003eMaturana MI, Meisel C, Dell K, Karoly PJ, D\u0026rsquo;Souza W, Grayden DB, et al. Critical slowing down as a biomarker for seizure susceptibility. Nat Commun. 2020;11:2172. doi:10.1038/s41467-020-15908-3\u003c/li\u003e\n\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":true,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"Parkinson’s disease, freezing of gait, EEG, closed-loop stimulation, early warning","lastPublishedDoi":"10.21203/rs.3.rs-7814436/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-7814436/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003e\u003cstrong\u003eObjective: \u003c/strong\u003eTo investigate whether scalp electroencephalographic (EEG) early-warning markers can predict freezing of gait (FoG) in Parkinson’s disease (PD) and to evaluate the therapeutic effects of a non-invasive closed-loop neuromodulation approach.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eMethods: \u003c/strong\u003eWe retrospectively reviewed 100 PD patients with clinically confirmed FoG. Patients with prior deep brain stimulation (DBS) were excluded. Based on treatment records, patients were allocated to a closed-loop group (n=50) receiving EEG-informed transcranial alternating current stimulation (tACS) or a control group (n=50) receiving optimized medical therapy and physiotherapy. EEG variance, lag-1 autocorrelation, beta-band synchronization, and network entropy were extracted as early-warning signals. Outcomes included FoG frequency and duration, gait velocity, stride variability, UPDRS-III gait item, and FoG-Q, assessed at baseline, 2, 4, 8 weeks, and 3 months.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eResults: \u003c/strong\u003eCompared with controls, patients receiving closed-loop tACS showed significant improvements from Week 4 onward, including fewer and shorter FoG episodes, faster gait, and reduced stride variability (all p \u0026lt; 0.001 at 3 months). Clinical gains were paralleled by EEG changes, with reductions in variance, lag-1 autocorrelation, and beta synchrony, and increases in network entropy, all correlating with FoG improvement.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConclusion: \u003c/strong\u003eNon-invasive closed-loop tACS guided by EEG early-warning markers significantly alleviates FoG and improves gait stability in PD patients. These findings highlight variance, lag-1 autocorrelation, and beta synchrony as practical biomarkers for personalized neuromodulation. Longer-term studies are needed to establish durability and safety.\u003c/p\u003e","manuscriptTitle":"Identifying Neural Tipping Points of Freezing of Gait in Parkinson’s Disease: An EEG-Based Early Warning and Closed-Loop Stimulation Study","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-11-14 10:08:35","doi":"10.21203/rs.3.rs-7814436/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"a7b02483-5d15-48d7-943f-707f0e51ae05","owner":[],"postedDate":"November 14th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[{"id":57887871,"name":"Health sciences/Biomarkers"},{"id":57887872,"name":"Health sciences/Neurology"},{"id":57887873,"name":"Biological sciences/Neuroscience"}],"tags":[],"updatedAt":"2025-11-20T12:53:52+00:00","versionOfRecord":[],"versionCreatedAt":"2025-11-14 10:08:35","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-7814436","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-7814436","identity":"rs-7814436","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}