The primary objective of this single-center, no-profit, longitudinal interventional randomized controlled, single-blind trial is to compare the effects of 2 different treadmill training treatments using C-Mill: the experimental one, endowed with augmented virtual reality (AVR) applications, versus the conventional one, the standard treadmill training in PD patients with gait and or balance disturbances. The main questions the study aims to answer are 1) Is the experimental treatment more effective than the conventional one? 2) Is it possible to identify predictive and indicative biomarkers of an outcome measure of rehabilitation using extracellular vesicles (cEVs) assessed by Raman spectroscopy? Participants will be randomized into two groups: the experimental group that will receive the experimental intervention, and the control group that will receive the conventional intervention. Both groups will train three times per week for 8 weeks, the first session starting from 25 minutes (25'). The experimental and the conventional treatments are planned to be progressive and will be individualized to the participant's level of performance.

Clinical, neuropsychological, and instrumental variables will be collected at baseline (T0), at the end of the treatment (T1), and 3 months after the end of treatment (T2). At 6 months after the end of treatment (T3), a phone interview will be performed. Both within-group and between-group analyses will be conducted. Biosamples will be collected at baseline (T0) and at the end of treatment (T1).

Study design: single-center, no-profit, randomized controlled single-blind trial (clinicians that assessed the effect of the interventions will be blind), with an active comparator. The study reflects the design of superiority of the experimental treatment versus the standard one. The treatment arm will be assigned by randomization. The results obtained with the 2 treatments will be compared.

The sponsor of the study is the Department of Clinical and Experimental Medicine, University of Florence, Italy; the study will be performed at "Struttura Organizzativa Dipartimentale (SOD) di Riabilitazione Generale- Istituto di Ricovero e Cura a Carattere Scientifico (IRCCS) Fondazione Don Carlo Gnocchi, Florence".

Collaborators for external services:

• Movement Analysis Laboratory at IRCCS Don Carlo Gnocchi Florence
• Laboratory of Nanomedicine and Clinical Biophotonics (LABION) at IRCCS Don Carlo Gnocchi Milan

Study start and length: study started on 6 July 2022 and will terminate on December 2023.

Study length for single participant 8 months: enrollment and baseline assessment (T0), treatment (2 months duration), end-of-treatment assessment (T1), follow-up assessment at 3 months after the end of treatment (T2), telephone interview at 6 months after the end of treatment.

Primary objective: to compare the effects of an integrated rehabilitation intervention using a new treadmill training that includes AVR applications, versus a conventional treadmill-based intervention in patients with PD (Hoehn and Yahr stage II-III) affected by walking and or balance disorders.

Secondary objective: to stratify PD patients based on their biological profile and identify predictive biomarkers and biomarkers indicative of an outcome measure of rehabilitation.

Description of the intervention. The intervention includes three sessions per week for 8 weeks. The intervention will be delivered during the "on" time of patients, every day at the same time.

The experimental and the conventional treatments are planned to be progressive and customizable to the participant's level of performance:

• progression in gait speed: gait speed is set at 80% of the individual's overground walking speed at the beginning of training, and will be weekly progressively increased to a maximum of 120%
• progression in trial duration: at the beginning of training, the duration is set at 25 minutes, including 5 slots of exercises lasting 5 minutes, with a 4-minute rest between each slot; every two weeks it will be increased by 1 minute per slot, reaching a maximal duration of 45 minutes in the last two weeks of treatment;
• progression in trial difficulty (only for the experimental group): the protocol has 5 difficulty levels for each C-Mill application, and the transition among levels is set at 80% success achievement.

Safety measures: participants will exercise wearing an anti-fall device and heart rate control; when the heart rate exceeds the safety threshold (set at 75% of HRmax, i.e., 220-age for males; 200-age for females), the treadmill speed will be lowered until the parameter is normalized.

Motek's C-Mill is a newly developed treadmill for gait and balance assessment and training that will be used for the study in both groups in a safe way (with an anti-fall device application and heart rate control). It is a treadmill sensor inclusive of a force platform. The patient is instructed to avoid using the side support bars during the training. The AVR applications of the C-Mill included in the training protocol are "nature island", "stepping stones", "walking area", "obstacles avoidance", and "tracks". These applications train balance and changes in walking speed, promote gait adaptation strategies and strategies to overcome freezing of gait, in a safe and controlled environment. Moreover, feedback to promote proper walking is provided (by physiotherapists and applications) including feedback on gait parameters such as stride symmetry, stride length, and cadence.

Visits planning and assessment timing Clinical, neuropsychological, and instrumental variables will be collected at baseline (T0), at the end of the treatment (T1), and 3 months after the end of treatment (T2). At 6 months after the end of treatment (T3) a phone interview will be performed.

All the assessments will be performed at each time point (T0, T1, T2) with some exceptions: automatic acquisition of gait parameters using C-Mill in C-Gait mode will be collected only at T1 and T2; at the end of treatment (T1), a 5-point Likert scale will also be administered to register patient satisfaction; during the phone interview at T3 only the falls questionnaire will be administered.

All visits will be planned in the "on" time of patients. Pharmacological treatments should be stable until T1.

Definition of adherence: adherence to the intervention will be considered sufficient if the patient respects the times and methods of execution of the rehabilitation treatment indicated in the intervention scheme. Missing up to 5 sessions is allowed, and lost sessions will be possibly recovered at the end of treatment. In case of discontinuation of treatment (one or more sessions missed), the treatment will be restarted with the gait speed, trial duration, and level of difficulty used in the last completed session. Participants who will miss more than 5 sessions will be considered dropouts.

Sample size. The G*Power software was used to estimate the sample size. From the literature, previous studies that aimed to evaluate the effects of treadmill training other than routine (partial weight-supported treadmill training) on patients with PD had observed a high effect size (ɳ2=0.737). In our estimate, a medium effect size (f=0.25) was conservatively chosen. Assuming a statistical power of 95% and an α=0.05, the resulting sample is 22 subjects per arm. To compensate for possible drop-outs, estimated at around 35%, the enrollment of a further 16 patients is appropriate, 8 per treatment arm, reaching an estimate of 30 subjects per group.

Data analysis. For all data, the distribution will be assessed using the Kolmogorov-Smirnov test (assuming the presence of normal distribution when p>0.05). Data will then be summarised as mean and standard deviation, median and interquartile range, or absolute and percentage frequency, as appropriate. The two groups will be compared at baseline to explore significant differences in clinical and demographic variables. Both within- and between-groups analysis will be conducted to assess the effects of the treatments delivered, for both primary and secondary outcome measures. Specifically, a repeated measures ANOVA will be used with a within-factor (Time of assessment) and a between-factor (Group). Statistical analysis will be conducted using the software International Business Machines Corporation (IBM) Statistical Package for the Social Sciences (SPSS) v28.

The acquired Raman spectra will be analyzed by multivariate Principal Component Analysis - Linear Discriminant Analysis (PCA-LDA) classification. PCA will reduce the number of variables into principal components, which will be used to build the LDA model. The model will be built to discriminate clinical improvement detected as quantifiable change on clinical scales, defined based on the minimum clinically important difference, when available from the literature. Sensitivity, specificity and accuracy of the predicting model based on spectra data will be assessed. In addition, correlation analysis will be conducted to evaluate the association between spectral data and change on clinical scales recorded between the beginning and the end of the treatment. Statistical analysis of the Raman data will be performed using Origin2021 software.

Univariate analysis models will be applied to select the best prognostic markers of clinical improvement to be included in the multivariate model. For all analyses, statistical significance will be set at p<0.05

Inclusion Criteria:

• Diagnosis of Parkinson's disease according to the diagnostic (POSTUMA criteria)
• Hoehn and Yahr Stage II-III
• Age>18 years
• One fall in the past 3 months/presence of postural instability /gait disturbance
• Able to walk for at least 5 minutes without assistance
• Stable drug therapy by at least 1 month
• Willingness to participate in the study, ability to understand and willingness to sign informed consent

Exclusion Criteria:

• Other pathology concurrent with gait disturbance (symptomatic arthritis involving hip/knee/ankle, stroke outcomes, severe polyneuropathy)
• Cognitive impairment capable of interfering with rehabilitation procedures, estimated as a score less than 18.58 at the Montreal Cognitive Assessment (MoCA), row score corrected according to Aiello et al, 2022
• Hallucinations
• Psychiatric disorder not controlled by current drug therapy
• Alcohol/drug use
• Uncompensated visual/auditory deficit that limits enjoyment of the cues provided by the AVR
• Communication deficit from any cause that impairs understanding of the task and the objectives of the intervention
• Recurrent episodes of severe orthostatic hypotension
• Severe cardiovascular diseases
• Patient undergoing other experimental protocol (patients regularly undergoing physical activity or sport will not be excluded)

Outcome Assessors will be blinded regarding treatment assignment, including clinicians assessing the eligibility of patients and enrolling them in the study; the evaluators (neurologist, physiatrist, psychologist, physical therapists, engineers, laboratory staff).

Blindness will be ensured by the modality and timing of assessments at T0, T1, and T2 that will be conducted at times other than treatment and by evaluators not involved in the treadmill training. Blinding will be removed if requested by the participant at the end of the study and in case of undesirable effects.