The trial was designed following a prospective, randomized, and controlled protocol. The investigation was conducted at a rehabilitation hospital focused on neurology patients and certified as a rehabilitation center for Parkinsonian diseases. Primarily, 87 potential patients were identified of which 74 were eventually included (Figure 1). Patients were included if they had a rehabilitation treatment of at least 3 weeks, had a diagnosis of at least one neurological movement disorder, were of legal age, and had capacity. Exclusion criteria were severe visual impairments, severe dementia, and inability to walk. During the course of the study, there was one instance of attrition where a patient was unable to complete the full 3-week protocol. This occurred because the patient experienced recurring syncopes and orthostatic instability unrelated to this trial, requiring an acute referral to another hospital for further medical intervention. As a result, the data from this particular case had to be excluded from the analysis to maintain the integrity and consistency. By removing the incomplete data from the analysis, the overall validity of the study’s findings can be preserved.

The selected patients were randomized using a matched pairs design dividing them into three groups: Intervention group 1 and Intervention Group 2 (IG1 and IG2) containing 25 and 24 patients, respectively, and a control group (CG) containing 25 patients. To ensure balanced groups and minimize potential bias, a merging process was implemented to assign patients to triplets based on their baseline characteristics at a rough estimate. Within each triplet patients were then randomly assigned to one of the three treatment groups helping to distribute any potential confounding factors equally among the groups and enhancing the validity of results. Baseline characteristics including age, number of patients with DBS system, Hoehn and Yahr score, duration of disease, and duration of rehabilitation treatment were similar between each group (Table 1). There was no further group stratification based on DBS.

The purpose of the study, the associated risks, potential outcomes, and the anonymized usage of data were thoroughly explained to the patients who were assigned. The informed consent process was conducted, and the patients provided their consent in written form, indicating their understanding and agreement to participate. It was also made clear that they had the right to refuse participation or withdraw from the trial at any point without the need to provide reasons. By providing comprehensive information and obtaining informed consent, the study adhered to ethical guidelines and ensured that the patients were well-informed participants in the research process.

An idealized training scheme was set up to distribute training days balanced throughout the total intervention time of 3 weeks. Meanwhile, all patients continued receiving standard rehabilitation physical therapy and medical optimization in terms of medication and non-medication assistance. Patients in the IG1 trained twice weekly, either on Mondays and Thursdays or on Tuesdays and Fridays resulting in a total of six training days. Patients in the IG2 trained every day within the week (Mondays to Fridays), thereby receiving 15 days of exercise in total. Patients in the CG were treated with conventional rehabilitation therapy only.

To exclude the adaption of medication as a disruptive factor for the interpretation of changes in the physical agility of patients, L-Dopa medication was assessed pre- and post-intervention. L-Dopa equivalent doses were calculated to ensure comparability. Following the introduction of safinamide in 2015 and opicapone in 2016, the previously used conversion (18) was extended (19) and utilized in this trial. Table 1 shows that all three groups increased their L-Dopa equivalent dose whereas there is no difference between each of them in both pre- and post-assessment. It was not feasible to conduct a more precise registration of On–off-Status and exact medication administration per day and per patient. Therefore, these specific details were not recorded or included in the study’s data collection process.

The inventory process and the intended purpose of this system were arranged in concordance with the latest suggestions by the MDS Task Force on Technology (20). While inventing the training system, primary body movement disorders such as gait and balance disorders, camptocormic posture and gait abnormalities, rigidity, akinesia, tremor, and fine motor skills disorders were identified, most of which were integrated into the conceptual planning. Specific movement patterns were defined which are to be practiced with the support of the therapy system and which counteract the above-mentioned disorders in a targeted manner. Particular attention was paid to stretching the upper body and getting the patient to stand up and sit down addressing greater walking and standing stability and the speed of movement. Another requirement for the system was the recognition of essential symptoms via marker-free sensor systems. Despite its approved usage for diagnostical purposes (21, 22), it was not to be assumed at the current time of processing, that resting, action and postural tremors of the hands could be recorded with the Microsoft® Kinect sensor system. To avoid fatigue and other adverse reactions, the duration of the training had to be adjusted accordingly and break times were considered. Cognitive exercises are integrated into the system to bridge them.

The established training system is based on the markerless sensing Microsoft Kinect® camera which can easily be connected to any computer system and runs with the developed training software without additional software needed. Moreover, the setup requires a simple LCD monitor such as a TV set that most people possess in their homes. For the training session, the patient has to stand in front of the camera at a distance of two to four meters and needs enough space to move their arms freely. No additional software or hardware is needed which makes the system very feasible and safe to use. The software includes guiding instructions for the whole game, as well as previews for all the movements the patients have to perform within the different games making it completely self-explanatory. Symptoms of Parkinsonian diseases tend to progress with time showing a successive decline in movement amplitude and speed (23). To counteract these developments the games were designed to condition a faster and wider sequence of movement. Similar concepts of progressive training have been used by Vieira de Moraes Filho et al. (24) eventually improving brady- and akinesia over time. In an overall game time of approximately 20 min, patients play four mini-games (Figure 2) with a focus on upper limb movement and the rise from a chair targeting the extension of the range and speed of motion. Table 2, therefore, comprises the intended purposes for every specific movement trained by one of the four exercises. Additionally, the software records the height of the avatar and the time needed to trigger each following object within the different games to assess the progress of the player afterwards.

Clinical examinations and measurements included in this study were made by the conducting staff and by well-trained physiotherapists. To objectify the findings, the change in the Unified Parkinson’s Disease Rating Scale (part III) of The International Parkinson and Movement Disorders Society (MDS-UPDRS-III) was used as the primary outcome variable. This scale is well established for the assessment of motoric symptoms of patients with Parkinsonian diseases (25). The data collection was conducted following an idealized schedule with an observation period of 3 weeks (21 days) for every patient. Clinical measurements and MDS-UPDRS-III were taken before the first and after the last training session of the interventional groups and on the first and last day of the three-week study term of the CG, respectively. The assessments mentioned are routine procedures conducted during admissions to the rehabilitation center where the trial took place. To ensure the accuracy and consistency of the assessments, the team of physiotherapists responsible for conducting the UPDRS-III assessments received comprehensive training from experienced neurologists. As a result, they were highly proficient in administering the tests and were well-versed in the evaluation process. The UPDRS-III assessments were performed on an individual basis and in a single-blinded manner. This means that the conducting physiotherapists were unaware of both the training status and group assignment of each patient. This approach helped minimize potential bias and ensured the objectivity of the assessments.

To extend the evaluation to record elusive changes in mobility that are not covered by the findings of the MDS-UPDRS-III assessment, further clinical measurements and tests were applied. The patients were asked to stand straight and stretch out their arms as far above their heads as possible. Then, the distance from the fingertips to the floor was measured to investigate the maximal erecting of the body. This is meant to address the ability to reach out to objects that are placed overhead which is a very important skill in day-to-day life. Apart from that, the greatest abduction angle of both arms was documented by taking photographs and was later quantified using the graphical software GIMP®. The same software was used to evaluate the angle of camptocormia as a marker for the severity of the posture impairment. Additionally, patients had to do a 5-step-walking test which primarily focused on freezing symptoms after the initial “start” command and which was meant to assess gait impairments. Therefore, time and distance were recorded. Due to the nature of the study and the involvement of the main researcher in both the training sessions and data collection, maintaining blinding for data acquisition of these additional measurements was not feasible. However, efforts were made to ensure objectivity and consistency in the data collection process.

Statistical analysis was conducted using IBM SPSS Statistics 25.0® (Armonk, United States). The normal distribution of the outcome parameters was assessed using the Shapiro–Wilk test, given the small sample size. Depending on the nature of the data, either parametric or non-parametric test procedures were employed. Specifically, the Mann–Whitney U-test and Kruskal-Wallis test were used for continuous variables, while the Pearson’s Chi² test was applied for categorical variables. A significance level of 0.05 was used for all test procedures.