Simulating space walking: a systematic review on anti-gravity technology in neurorehabilitation

Abstract

Neurological disorders, such as Parkinson’s disease (PD), multiple sclerosis (MS), cerebral palsy (CP) and stroke are well-known causes of gait and balance alterations. Innovative devices (i.e., robotics) are often used to promote motor recovery. As an alternative, anti-gravity treadmills, which were developed by NASA, allow early mobilization, walking with less effort to reduce gait energy costs and fatigue. A systematic search, according to PRISMA guidelines, was conducted for all peer-reviewed articles published from January 2010 through September 2023, using the following databases: PubMed, Scopus, PEDro and IEEE Xplore. After an accurate screening, we selected only 16 articles (e.g., 5 RCTs, 2 clinical trials, 7 pilot studies, 1 prospective study and 1 exploratory study). The evidence collected in this systematic review reported promising results in the field of anti-gravity technology for neurological patients, in terms of improvement in gait and balance outcomes. However, we are not able to provide any clinical recommendation about the dose and parameters of anti-gravity treadmill training, because of the lack of robust high-quality RCT studies and large samples.

Registration number CRD42023459665.

Introduction

Neurological disorders, such as Parkinson’s disease (PD), multiple sclerosis (MS), cerebral palsy (CP) and stroke are well-known causes of gait and balance alterations [1]. Reduced mobility owing to neurological disorders is associated with multiple consequences on cardio-vascular and muscle-skeletal systems, limiting activities of daily living and patients’ quality of life. In this context, innovative devices (i.e., robotics) are exploited in the neurorehabilitation field [2, 3]. In fact, robotic devices (such as exoskeletons and end-effectors) facilitate walking functions, even in patients with severe motor deficits due to brain damage [4]. However, these systems can limit joints movement due to the constraint of the robotic orthosis and may not allow normal gait patterns. As an alternative, NASA researchers have developed a new technology that mimics antigravity and uses differential air pressure to train astronauts to counteract muscle and bone loss. This technology consists of anti-gravity treadmills (A-GT), in which the lower half of the subject is surrounded by an air-tight, enclosed inflatable bag [5]. When the air compressor reaches the pressure in the chamber above atmospheric pressure, it creates an axial buoyant force, allowing gait training. Specifically, the air is released after the subject’s weight calibration and the calibrated weight is used as a reference for selected unweighting during exercise [5]. In addition, the anti-gravity treadmills can be used by participants of all heights, thanks to vertical frame height adjustment. The body weight support system can sustain 80% of a person’s body weight and can be adjusted progressively [6]. The safety and feasibility of A-GT was already investigated in healthy subjects, as well as in orthopaedics, post-surgical patients, and in neurological disorders [5, 7, 8]. The potential benefits of using A-GT in a neurorehabilitation context include early mobilization, walking with less effort to reduce gait energy costs and fatigue, decreasing the harmful impact on injured joints and maintain cardiorespiratory fitness [8]. One of the most used A-GTs in neurorehabilitation is the Alter G (AlterG Sports, AlterG Inc., California, USA). This helps to maintain normal muscle activation and gait patterns [9]. Thus, the use of A-GT could be an adjunctive rehabilitation treatment in those neurological patients who may manifest moderate motor deficits, allowing long-lasting aerobic training to promote neuroplastic processes. It is noteworthy that the use of A-GT could particularly involve vestibular pathways, reinforcing sensory and proprioceptive feedback, thus activating cortical areas (e.g., primary somatosensory cortex, motor cortex, insula, parietal and occipital lobes and frontal areas) [10]. In addition, aerobic exercise is a well-known way to improve neuroplasticity, as it promotes the release of neurotrophic factors like brain-derived neurotrophic factor (BDNF) [11]. However, it is still unclear whether A-GT could be beneficial and/or effective as an adjunctive innovative treatment in neurological patients. The main objective of this systematic review is to investigate the literature about the effects and potential benefits of A-GT training in neurological disorders, including PD, SM, CP, and stroke. These conditions collectively represent a significant proportion of neurological disorders worldwide and are associated with substantial gait impairment that is not easy to manage with conventional physiotherapy alone.

Methods

The protocol of this systematic review was registered on PROSPERO (https://www.crd.york.ac.uk/prospero) with the registration number CRD42023459665, following the Preferred Reporting Items for Systematic and Meta-analyses (PRISMA) [12]. Our research is aimed to explore the existing evidence on the effects and potential benefits of anti-gravity technologies in the context of neurorehabilitation.

PICO model

Search terms were defined according to PICO model (Population, Intervention, Comparison, Outcome) [13]. The population included patients affected by neurological disorders, such as stroke, PD, CP and MS; intervention included all anti-gravity existing technologies in the field of neurorehabilitation; the comparison included sham or placebo treatments, and/or conventional physiotherapy conducted in the control group, allowing for a comparative analysis of the effects of the active interventions. However, considering the limited literature available, we included multiple study designs for qualitative synthesis, such as non-controlled/randomised studies; and outcomes included any motor improvements shown by the patients and efficacy of treatment.

Search strategy and eligibility criteria

A systematic search, according to PRISMA guidelines [12] (see Supplementary material for PRISMA checklist), was conducted for all peer-reviewed articles published from January 2010 through September 2023, in order to search for the most recent literature. We chose to include articles from 2010 because of the growing interest in technology in neurorehabilitation. Our research was conducted on the following databases: PubMed, Embase, Cochrane Database, PEDro, Web of Science and IEEE Xplore. The following terms were used: (“neurological disorders”) OR (“stroke”) OR (“Parkinson’s disease”) OR (“multiple sclerosis”) OR (“cerebral palsy) AND (“anti-gravity technology”) OR (“anti-gravity treadmill”) OR (“Alter G”).

All articles were reviewed based on titles and abstracts by two investigators (M.B and R.S.C), who independently performed data collection to reduce the risk of bias (i.e., the bias of missing results). These researchers read the full-text articles deemed suitable for the study and in case of disagreement on the inclusion and exclusion criteria, the final decision was made by a third researcher (M.G.M). The inclusion criteria were: (1) patients with neurological disorders due to central nervous system impairment, including stroke, Parkinson’s disease, multiple sclerosis, and cerebral palsy, since they collectively represent a significant proportion of neurological conditions worldwide and are associated with substantial gait impairment that it is not easy to manage only with conventional physiotherapy; (2) an applied approach to motor rehabilitation; (3) written in English; and (4) published in a peer-reviewed journal. We have excluded articles describing theoretical models, methodological approaches, algorithms, and basic technical descriptions. Additionally, we excluded: (1) animal studies; (2 conference proceedings or reviews; and (3) studies involving children affected by neurological disorders other than CP; (4) studies involving other neurological disorders that do not involve the central nervous system; (5) case reports and reviews. Our search strategy included some filters such as temporal range between 2010 and 2023. The searches were limited to the title and abstract in this phase. Additionally, we considered the reference lists of included papers for the screening to identify additional relevant papers not found by the search strategy. The list of articles was then refined for relevance, revised, and summarized, with the key topics identified from the summary based on the inclusion/exclusion criteria.

Data extraction and analysis

After full-text selection, the data extraction from the included studies was summarized in a table (Microsoft Excel – Version 2021). Data summarized were considered for the following information: assigned ID number, title of study, year of publication or presentation and first author, study aims and design, study duration, method and setting of recruitment, inclusion/exclusion criteria, use of a control group, use of devices, informed consent, conflict of interest and funding, type of intervention and control, number of participants, characteristics at the baseline, setting of intervention, type of outcome and time-points for assessment, adverse events, results and key conclusions. In addition, the agreement between the two reviewers (MB and MGM) was calculated through the kappa’s score [14]. The kappa score, which establishes a threshold for substantial agreement at > 0.81, was interpreted as reflecting excellent concordance between the reviewers. This criterion ensures a robust evaluation of inter-rater reliability, emphasizing the achievement of a substantial level of agreement in the data extraction process.

Data quality assessment

The quality of each article was rated by the two reviewers (MB and MGM) using a revised Cochrane risk of bias (RoB 2) [15] for 5 RCT studies [17,18,19,20,21]. RoB-2 consists of five domains: (i) bias arising from the randomization process, (ii) bias due to deviations from intended intervention, (iii) bias due to missing outcome data, (iv) bias in the measurement of the outcome, (v) bias in the selection of the reported result.

Moreover, we used ROBINS-I [16] for the other non-randomized studies [22,23,24,25,26,27,28,29,30,31,32]. ROBINS-I is a method used to assess the risk of bias in non-randomized research. This assessment tool considers seven domains of potential sources of bias: (i) bias due to unknown or uncontrolled confounding factors; (ii) bias due to selection of participants, (iii) bias due to classification of interventions, iii) bias due to measurement of variables, (iv) bias due to deviation from intended intervention; (v) bias due to missing data; iv) bias due to selection of measurement outcome; (vi) bias due to selection of reported results.

Synthesis of evidence

Our initial research revealed 240 results, then we excluded 120 articles due to eligibility criteria that were not fully respected. Finally, we removed duplicates and we included and analysed 16 articles dealing with A-GT in neurological disorders (see Fig. 1).

Fig. 1

PRISMA flow diagram

In details, we found 5 RCTs [17,18,19,20,21], 2 clinical trials [22, 23], 7 pilot studies [26,27,28,29,30,31,32], 1 prospective study [24] and 1 exploratory study [25]. Study populations of the included evidence range between 6 (the minimum) and 50 (the maximum) subjects affected by different neurological disorders. With regard to aetiology, we found: 5 articles (2 clinical trials and 3 pilot studies) about PD patients [22, 23, 26,27,28]; 1 pilot study on multiple sclerosis [29]; 7 articles on stroke patients [17,18,19,20,21, 24, 25], (5 RCTs, 1 exploratory study and 1 prospective study); 3 pilot studies [30,31,32] on children affected by CP (see Table 1).

Table 1 Description of study design, sample size and aetiology of neurological disorders

Quality of the studies and risk of bias

We found a great heterogeneity among the included studies that could influence the interpretation of results. Firstly, our research identified various study design, such as RCTs [17,18,19,20,21], clinical trials [22, 23], and pilot studies [26,27,28,29,30,31,32]. Most of the studies did not specify randomization procedures or blinding of raters, due to their methodology (e.g., pilot, exploratory, and prospective studies). Indeed, seven studies [22, 23, 26] did not include controls in their study design, which affected the assessment of A-GT intervention effects. Secondly, outcome measures varied among studies; some authors administered clinical tests/scales, including Unified Parkinson’s Disease Rating Scale (UPDRS) [26], Berg Balance Scale (BBS) [18, 19, 21], Tinetti POMA (Performance Oriented Mobility Assessment) [17, 18, 27], Balance evaluation systema test (BESTest) [32], 6-Minutes Walking Test (6MWT) [20–21,25,28 ], Functional Ambulation Classification (FAC) [17, 20, 24, 25], Timed Up and Go (TUG) [18, 19, 26, 28], 10-Metres Walking Test (10MWT) [18,2528] to assess both gait and balance functions. Only a few studies performed specific assessment procedures such as gait analysis [20, 30,31,32], electromyography (EMG) [22], near-infrared spectroscopy (NIRS) [29]. In addition, we found substantial variability in training protocols and settings of the selected studies, with even 3 studies [19,20,21] not explicitly describing the training protocol used. Most authors administered AG-T training alone [17,18,19, 21,22,23,24,25,26,27,28,29,30,31,32], apart from Sukonthamarn et al. [20], who combined conventional physiotherapy with A-GT. Lastly, the analysed papers reported a variable number of participants most of whom were in small sample sizes [24, 25, 28,29,30, 32]. Furthermore, we performed a risk of bias assessment via RoB 2 [15] in 5 RCTs [17,18,19,20,21]. The included studies showed good overall quality, except in one study [19] where the result was “some concerns” following lack of blinding (domain 2). Otherwise, a low risk emerged, particularly in domains 1, 4, and 5 across all studies (see Fig. 2).

Fig. 2

RoB 2 assessment of the 5 RCT studies

Risk of bias for the remaining eleven non-RCTs [22,23,24,25,26,27,28,29,30,31,32] was performed by using ROBINS-I [16], as reported in Fig. 3. We noticed that most of the included studies reported a moderate risk of bias, especially in the domains 2 [22, 23, 27, 28, 30, 32],4 [23,24,25,26,27,28,29, 32], 5 [22,23,24,25, 27,28,29, 31, 32], and 7 [22,23,24,25,26,27,28,29, 32]. In the majority of the studies, inclusion/exclusion criteria of the participants were not fully reported, or they were not explained clearly. In addition, the participants in the included studies reported low adherence to the treatment, even if adverse events were not reported. With regard to domains 5 and 7, we noticed that some concerns were present related to participants missing data, and to multiple analyses with small samples that can increase the overall risk of bias (see Fig. 4).

Fig. 3

ROBINS-I assessment of the eleven non-RCTs studies

Fig. 4

ROBINS-I assessment of the eleven non-RCTs studies

Description of intervention

All studies included in this systematic review administered A-GT training, comparing it with conventional treadmill, aquatic treadmill and/or conventional gait training/exercises. The A-GT training lasted from 20 to 60 min per session in each study, considering patients’ tolerance. Some authors [24, 25, 29] also performed a 20 to 30 min warm-up before A-GT session in order to prepare patients for further aerobic effort. The training periods reported by the authors ranged between 4 and 8 weeks, although Rigby et al. [27] performed the longest training period which lasted for 24 weeks. In general, the anti-gravity support during training was estimated with patients’ body weight, and it was slowly increased according to patients’ needs [17, 18, 23,24,25,26,27,28, 30,31,32]. However, there is a substantial heterogeneity of A-GT training protocols among the selected studies, as reported in Table 2. For example, Malling et al. [23] performed a specific training protocol in three blocks: motor, aerobic and mixed, using the A-GT. Furthermore, we found other differences among the included articles related to the % of unloading body weight support. Most of the authors [24,25,26,27,28] considered 50% of body weight unloading, which was gradually adjusted, according to the patients’ needs. Specifically, Almutairi et al. [24, 25] performed a training protocol starting from 50% of unloading, which was gradually decreased by 2% in each session. In contrast, Oh et al. [18], considered an initial 30% of overload, which was gradually increased to 80% of the patient’s body weight.

Table 2 Description of anti-gravity treadmill training parameters and settings

Effects of intervention

All the included studies [17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32] investigated the role of the A-GT in improving endurance, balance and gait functions. Authors administered A-GT training using the AlterG (Inc., California, USA) for the experimental procedures. In the control groups, six studies used conventional physiotherapy [17,18,19,20,21, 24, 25, 30], while three studies used gait training with other types of land and/or aquatic treadmills [27, 28, 31]. However, seven studies [22,23,24,25,26, 29, 32] out of the 16 [17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32], did not include a control group in their study design. The outcome measures included were mainly oriented towards measuring gait [17,18,19,20,21,22, 26,27,28,29,30,31,32] and balance [18, 21, 23, 24, 27, 28] functions, whereas cardiovascular function (e.g., heart rate, blood pressure and oxygen saturation) was assessed in three studies [21, 24, 28]. Additionally, Willingham et al. [29] evaluated muscle oxidative metabolism and endurance through NIRS and mechanomyography in people affected by MS. Another study [22] tested force and EMG signals in the lower limbs of PD patients, during A-GT training, whereas gait analysis was performed only by Sukonthamarn et al. [20] and Aras et al. [31] (see Table 3).

Table 3 Description of the reported interventions, outcomes and major findings

Legend: TUG (Timed UP and Go), UPDRS-III (Unified Parkinson’s Disease Rating Scale), Tinetti POMA (Performance Oriented Mobility Assessment), 10MWT (10-Metres Walking Test), 6-MWT (6-Minutes Walking Test), FAC (Functional Ambulation Classification), TS (Tinetti Scale), HR (Heart rate), BBS (Berg Balance scale), MMSE (Mini-mental status examination), RATE (Robotic assisted treadmill exercise), ATE (Antigravity treadmill exercise), BESTest (Balance evaluation system test), NIRS (Near-infrared spectroscopy).

Moreover, the effects of A-GT training included improvements in global mobility, freezing of gait (FoG) [26, 28], balance [23, 28] and gait functions [22] in patients with PD. Willingham et al. [29], reported promising results on endurance and muscle oxidative state in individuals affected with moderate to severe MS. In post-stroke patients, authors reported improvements in gait speed, endurance [19, 24, 25], balance and walking functions [19, 20] and cardio-respiratory fitness [21]. However, balance functions measured with POMA score was less statistically significant than POMA gait score, as showed by Oh et al. [26]. In post-stroke patients, specific improvements in lower limb muscle activation were also achieved after A-GT, as demonstrated by Calabrò et al. [17]. The EMG data, analysed as root mean squares, showed that the activity levels of the gastrocnemius and rectus femoris muscles decreased during the early and mid-swing phases of the gait cycle, specifically at 50%, 60%, and 70% of the cycle. In contrast, the activity of the tibialis anterior on the unaffected side increased during the preparation for heel strike, at 100% of the gait cycle. [17]. Positive results were also documented in CP patients, especially for balance, gait and risk of falling [30,31,32]. (see Fig. 3).

Discussion

To the best of our knowledge, this is one of the few systematic reviews [8, 33] investigating the effects of A-GT training in patients affected by neurological disorders. We have found that few articles dealt with this topic and most of them present high risk of bias [22,23,24,25,26,27,28,29, 32] making interpretation and generalization of results difficult. In particular, the novelty of our systematic review is the investigation of literature about the clinical effects of using anti-gravity technology in the neurorehabilitation context. Other authors have previously addressed our topic in different patient populations, such as paediatric [8] and orthopaedic (i.e., lower limb surgery and athletes’ injuries) [5,6,7]. Recently, a systematic review [33] highlighted the lack of larger RCTs and standardized training protocols using lower body positive pressure treadmills like Alter-G in neurological patients.

However, the authors did not include any articles about CP patients and did not perform a quality assessment or a risk of bias analysis for the included studies. These limitations should be considered when interpreting the results. In addition, Almutairi [33] considered also case reports in his analysis, which we excluded a priori due to their scarce scientific reproducibility. Moreover, we reported some clinical implications of using Alter-G in both acquired brain injury (i.e., stroke and cerebral palsy) and neurodegenerative disorders (i.e., PD and MS).

Acquired brain injury

In this paragraph, we discuss our findings related to acquired brain injury, including post-stroke and CP patients. Balance and gait disorders are the most common deficits in patients affected by neurological disorders and the prognosis to regain ambulatory functions depends on the underlying pathology and its severity [34]. Generally, robotic device like end-effectors and exoskeletons are often used in rehabilitation after neurological damage, including stroke and CP [35]. According to Calabrò et al. [36], the use of robotic gait training in post-stroke patients increases the possibility of regaining an independent gait, and this should be considered as either “add on” treatment or even in substitution of the traditional rehabilitation. Otherwise, the treadmill training in post-stroke patients appeared to be less effective in improving gait distance and balance when compared with overground gait training, as suggested by Gelaw et al. [37]. Moreover, Bonanno et al. [38], investigated the effects of robotic gait training on walk and balance functions in CP. The authors found that CP children with more severe disability may benefit from exoskeletons (since they have better joint and trunk control), whereas less impaired CP children may be trained with end-effectors and VR devices (as they require spared motor function). As an alternative, the A-GT, like Alter-G, allows a body weight supported gait, thus maintaining normal gait patterns [39]. In this sense, the Alter-G mostly improved temporal parameters of gait, such as gait speed in both post-stroke [18, 25] and CP patients [30, 32]. Hence, Calabrò et al. [17] demonstrated that the A-GT training in post-stroke patients shaped biceps femoris and rectus femoris bilaterally, which are essential muscles in opposing gravity force. Notably, the activation of rectus femoris allows propulsive forces during stance phase and this may lead the improvements in temporal-spatial variables of gait [40, 41]. Furthermore, as suggested by some authors [42] the A-GT could improve neuroplastic processes in the brain stem and cerebellar white matter after the training. In fact, these brain areas are fundamental in postural control and motor learning. Specifically, the cerebellum plays a role in modulating the step cycle to adjust step patterns, whereas the basal ganglia and the frontal cortex are involved in regulating gait during rapid changes in environmental conditions [43]. Azizi et al. [10] showed that A-GT training may lead to improvement in neurophysiological (motor evoked potentials-MEPs) and neuroimaging (diffusion tension imaging-DTI) indices of the corticospinal and vestibulospinal tracts in CP children. The idea is that the Alter-G, through a micro-gravity environment, could boost high myelination, improving balance and gait abilities [8,9,10, 44]. However, the investigation of brain activation related to weight-supported walking remains a challenging question.

Neurodegenerative disorders

Neurodegenerative disorders, like PD and MS, can cause progressive neuronal loss that consequently worsens postural control and gait ability over time [45]. Generally, conventional rehabilitation approaches include aerobic treadmill training, core exercises to improve balance reactions and postural stability, and hydrotherapy to reduce muscle stiffness and improve gait function [46]. Moreover, combined physiotherapy exercise training (including aerobic, resistance, and balance training) has shown beneficial effects not only for balance, muscle strength, gait recovery, and endurance, but also for slowing the progression of motor impairments [47]. In our systematic review, we noticed that literature about the use of A-GT in PD patients mostly improved global motor functions, reducing tremors, and freezing of gait. In addition, the A-GT training could improve kinematic factors of lower limbs, as suggested by Rose et al. [22]. The authors found that an eight-weeks A-GT training in patients affected by PD can normalize the extensor muscle activation during weight-supported gait. In line with these assumptions, Malaya et al. [9] showed that healthy subjects performing Alter-G training showed EMG elicited responses in the medial gastrocnemius as well as in the rectus femoris, which are both involved in the lower limb extension during gait. Furthermore, Berra et al. [48] compared the effects of treadmill training plus body weight support system with overground gait training. They found that the reduced body loading during gait training was effective in improving global motor skills and functioning, measured with UPDRS. However, they suggested that both types of gait training can be considered effective at inducing improvements in kinematic gait parameters. Moreover, A-GT seems to have a role in inducing muscle metabolic and plastic changes, as suggested by Willingham et al. [29]. In fact, the authors found that A-GT in MS patients improved muscle oxidative capacity through the activation of biochemical pathways, which are required for mitochondrial biogenesis. It is noteworthy that the study of Willingham et al. [29] is the only one which investigated the effects of A-GT in patients with MS. However, recent studies conducted with MS patients, using a treadmill, robotic devices, and partial body weight support systems, demonstrated the effectiveness of such training on in improving gait functions and global mobility [49, 50]. Lastly, the aerobic exercise is a well-known neuroplastic promoter [51]. Recent evidence suggests a strict relationship between cardiovascular performance and brain plasticity. It seems that intensive aerobic training is related to increased volume in hippocampus and basal ganglia which are involved in the control of motor behaviour [52, 53]. Altogether, these issues may explain the promising results of the selected studies in improving gait parameters and related functions.

Implications for clinical practice and future perspectives

To summarise, although both acquired brain injuries and neurodegenerative diseases can lead to motor impairment, the mechanisms of injury, disease progression, and motor recovery strategies can be very different. The potential benefits of AlterG have been studied primarily in acquired brain injury (i.e., post-stroke and CP) and in PD, a neurodegenerative disorder [17,18,19,20,21,22,23,24,25,26,27,28, 30,31,32]. From a clinical perspective, both post-stroke and CP patients achieved better outcomes in temporal parameters of gait (i.e. walking speed and cadence) and cardiovascular function [17,18,19,20,21, 24, 25, 30,31,32]. In this sense, AlterG may provide a safe and controlled environment to practise walking, which could explain the improvements in gait speed and cardiac fitness. Similarly, AlterG has shown potential benefits in gait function in CP children, likely due to the reduction in gravity that allows these children to practise walking with less effort [30,31,32]. On the other hand, improvements in global mobility (UPDRS), fall risk (TUG and POMA), freezing of gait and tremor have been obtained in PD patients [22, 23, 26,27,28]. These outcomes fundamentally differ from those observed in individuals with acquired brain injury, owing to the distinct pattern of neurological impairment. Despite these evident differences, both neurodegenerative conditions and acquired brain injuries exhibit the potential for enhancing overall walking functionality through the implementation of A-GT. Nonetheless, critical inquiries persist concerning the scarcity of RCTs with larger participant cohorts. Specifically, elucidating the optimal disease stage for initiating such interventions and determining the appropriate treatment dosage would be advantageous. Furthermore, many investigations have predominantly focused on gait kinematics (i.e., spatial-temporal parameters), with limited attention paid to EMG analyses [17, 22], while kinetic data, including forces and joint range of motion, have been largely overlooked. Finally, elucidating the activation patterns of cerebral regions (functional brain connectivity) during AlterG treatment would provide valuable insights. Comparing these patterns with those observed in alternative body weight support systems would further enhance our understanding.

Limitations

This systematic review has some limitations that need to be acknowledged. One limitation is the absence of quantitative analysis. In particular, we found considerable heterogeneity among the included studies in terms of methodologies, outcome measures, and participant characteristics, thus conducting a quantitative analysis was not feasible. This limitation underscores the need for caution in generalising the findings and emphasizes the importance of interpreting the results within the context of individual study characteristics. The selected studies also presented other limitations, including small sample sizes, lack of control group and lack of long-term follow-up evaluations. Despite these limitations, our review relies primarily on qualitative synthesis, based on systematically summarizing and interpreting the findings of individual studies to elucidate common themes, patterns, and discrepancies across the literature. As a result, our review provided a comprehensive qualitative synthesis of the available evidence, offering valuable insights into the novel A-GT rehabilitation approach for specific neurological conditions (i.e., PD, MS, CP and stroke), identifying key implications for clinical practice and considerations for future investigation.

Conclusion

The evidence collected in this systematic review shows promising results in the field of anti-gravity technology for neurological patients. When used alone or in combination with other treatments, the device can lead to better gait and balance parameters than conventional physiotherapy alone. However, we are unable to provide specific clinical recommendations about the dose and parameters of A-GT training, because of the lack of robust RCT studies and large samples. Future studies with rigorous methodologies should focus on comparing to other non-harness body-weight support systems, in order to better understand the potential effects of anti-gravity technologies.