An electronic database search was performed from 1st September to 31st December 2019 using the following databases: PubMed/MEDLINE, COCHRANE Library, PEDro, and SCOPUS. In line with the PRISMA guidelines, we then performed an additional manual search (e.g., through citations of articles included in this review). All databases were searched from the establishment of the database (January 2005) to 31 December 2019. The search strategy used a combination of MeSH (Medical Subject Headings) terms and keywords including “Parkinson's disease,” “Telerehabilitation,” “TR,” “Remote Rehabilitation,” “Home-based Rehabilitation,” and their related synonyms. The detail strategy is presented in Appendix 1. The PICOS principle (Population, Intervention, Comparison, Outcome measures, and Study design) was followed to define our research main question—Table 1.

Two authors (C.V., S.B.) conducted the literature search independently. First, they manually identified and excluded duplicate references, and then screened titles and abstracts for relevance. The eligibility criteria were agreed by consensus when comparing search results (C.V., S.B.). They were (i) the use of remote monitoring via computer devices and wearable sensors to treat PD patients at home, (ii) recording of human physical activity, (iii) adults aged 18–80 years with PD diagnosis, (iv) data published between January 2005 and December 2019, and (v) randomized controlled trial (RCT), review and systematic review, case study or series—Table 2. The exclusion criteria were (i) no use of wearable/portable sensors/TR system, (ii) management of other types of neurological dysfunction or musculoskeletal disorders, (iii) focus on cognitive deficit management with TR, (iv) article type = abstract, letter, poster or chapter from a book, (v) article not written in English, and (vi) full access to article not available—Table 3.

Data were extracted (C.V.) and checked (S.B., F.B., G.M.) with final adjudication by consensus. Variables extracted included the population sample studied, disease-specific severity levels, TR device name, intervention, setting, demographic data and study details, including design, funding sources, and motivational factors.

The search identified 689 articles: 258 were retrieved from PubMed Central, 85 from COCHRANE Library, 1 from PEDro, 304 from Science Direct, while 41 papers were identified by an additional manual search. After removing duplicate items (60), 629 articles remained. Of these, 320 were excluded based on the title only. Of the remaining 309 articles, a further 216 records were excluded after reading the abstract. We fully assessed the full texts of the remaining 93 papers according to the eligibility criteria, and 15 articles met the study criteria and were included in the final analysis (Figure 1). These 15 articles, involving 421 patients affected by PD, consisted of 8 RCTs (42, 55–61), 1 clinical trial (62), 2 case studies (63, 64), 3 pilot studies (65–67), and 1 study design (68).

The selected articles were divided into two categories based on their topic of interest or outcome.

The first category dealt with the effect of TR on gait and balance (3), dexterity of the upper limbs (3), and bradykinesia (0). The second category dealt with the effect of TR on non-motor symptoms such as speech disorders (8) and dysphagia (0). We also analyzed QoL (7) and patient satisfaction (8) with the TR program, as well as feasibility and costs—Table 4.

Only one study by Siegert et al. (68) investigated QoL as a primary outcome—Table 8. However, QoL was investigated in other studies whose main purpose was to evaluate the efficacy of TR on motor and non-motor disorders, respectively, gait and balance (42, 63), UL dexterity (56, 57), and speech and voice (60, 67). Change in QoL was analyzed through the Parkinson's Disease Questionnaire (PDQ-39) and improvement was shown in all these studies.

Specifically, PDQ-39 is a self-report questionnaire which assesses the perception of QoL in PD patients. Consists of 39 questions investigating difficulties in 8 domains such daily activities, physical discomfort, emotional and social aspect of the disease (71).

Only one study by Wilkinson et al. (61) investigated satisfaction after a telehealth program in people with PD as a primary outcome—Table 9. To assess patients satisfaction was used the Patient Assessment of Communication of Telehealth questionnaire (PACT), a 33-item validated questionnaire (Likert scale). Greater satisfaction with the telehealth modality was detected in the assessments of convenience and accessibility/distance.

Patient satisfaction was investigated in several studies as a secondary outcome. Research on motor (42, 63) and non-motor (58–60, 62, 64, 67) symptoms analyzed satisfaction after programs delivered by TR. In studies concerning motor aspects (gait and balance), the satisfaction was investigated using a 5-point Likert scale on a questionnaire in the RCT (42), while a semi-structured interview was conducted in the case study (63). A questionnaire with a nominal scale of 5 points was used on studies investigating non-motor aspect (speech) (58, 59, 64, 67); an opinion patients' was detected in Griffin's study (62). All studies reported high levels of agreement and patient satisfaction; only Gandolfi et al. (42) found no significant difference in satisfaction level.

Several different biases were detected that may affect the analysis of the results. Our search identified different types of studies: 8 RCTs (42, 55–61), 1 Clinical Trial (62), 2 case studies (63, 64), 3 pilot studies (65–67), and 1 design study (68). Most trials did not blind participants and outcome assessors [UL dexterity studies (56, 57) and patient satisfaction report (61)], while the RCTs considering gait and balance (42, 55) and speech and voice (58–60) were single-blinded examiner studies. Furthermore, the analyzed reports were carried out in a variable number of participants. Of these, although some subjects affected by PD were included, the heterogeneity of the sample was noted and the inclusion criteria were variable. The range of severity of PD score measured by the Hoehn and Yahr scale was often wide: the studies on motor symptoms presented a small variability [2.5–3 (42), 1–3 (55), 2–3 (56)] while those on speech and voice dysfunction showed a larger range [1–3 (67), 1–4 (58, 59), 1–5 (60)]. The other studies do not refer to the same scale. Moreover, some studies (42, 55, 56) assessed cognitive status by MMSE, while others (67, 68) utilized the MoCA. Some of the included studies had as primary outcome the feasibility and validity of TR treatment (55–60, 62–67), while only two studies (42, 57) investigating motor symptoms in PD had as primary endpoint some clinical outcomes. Finally, motor symptoms of PD were evaluated with different scales in the different studies—details are found in Tables 5–9. In contrast, for motor speech disorders we found a greater consensus about the methods and the scales used. In summary, it was difficult to perform a robust quality analysis for each category and the dataset was considered insufficient for the planned sensitivity analyses.

Although VR-based balance programs through TR have proved feasible and effective in several neurological conditions, in PD this system has been used separately to evaluate or to treat balance dysfunctions. Most studies related to VR application in patients with PD indicated that VR positively affected movement velocity, time, balance and gait, and postural control compared to healthy controls (79–81). Other studies that examined the effects of VR in PD patients reported a better improvement in postural stability and functioning compared to conventional treatments only (74, 82).

Others compared home-based VR training with conventional home-based balance training in patients with PD (83, 84); both groups showed similar improvements in balance and walking function. A few studies tested the Virtual Motor Rehabilitation (VMR) system, designed to improve postural control, in PD patients; preliminary results indicated improvement of balance, gait performance, and postural stability (8, 28, 81, 85, 86). To date, no study analyzed home-TR via VR systems specific for PD treatment. In fact, no specific systems have been designed. Exergames already on the market (i.e., Nintendo Wii) were used adapting them to the rehabilitation field. For example, Gandolfi et al. (42) compared improvements in postural stability after in-home VR-based balance training (TeleWii group) vs. inpatient sensory integration balance training (SIBT group). The TeleWii group performed a home-VR TR program consisting of graded exergames using the Nintendo Wii Fit system while the control group did inpatient SIBT including exercises to improve postural stability. The results suggested a similar improvement in balance (on the BBS) between the two groups.

On the contrary, studies on UL exergames found them to be acceptable and safe but they did not translate into improvement in functional activities. Tremor and rigidity in the UL can contribute to gross and fine motor coordination difficulties, which can subsequently adversely impact hand function (57). Exergames improved arm and hand activities with a home-based intervention (85). With a TR application, it was possible to perform a valid and reliable assessment of ADLs and hand function (56). Only the study by Cikajlo et al. (56) developed an intensive target-based physiotherapy for UL suitable for TR services (“FruitPicking” computer game). The TR services were comparable with usual treatment and the results showed a non-significant improvement in the dexterity of the hand.

To date only one paper (78) has evaluated the reliability and responsiveness of a motion sensor paired with a tablet app-based system for objective bradykinesia assessment both in the hospital and at home.

TR treatment studies for acquired neurologic speech disorders have mostly investigated the delivery of the LSVT LOUD. Individuals with PD are trained to “recalibrate” their motor and perceptual systems to improve self-monitoring and thus make more consistent use of the louder voice in daily communication (87). Various forms of synchronous and asynchronous technologies either as alternatives to, or in combination with, face-to-face delivery of LSVT LOUD have been investigated. The results indicate that internet-based assessment is generally reliable and valid. TR led to long-term improvements in vocal patterns and quality of voice and was well-accepted by patients with PD. The results showed high patient comfort and positive results in satisfaction. Patient perceptions have been explored (58–60, 64) using questionnaires following TR sessions. Most participants reported a positive experience and willingness to accept speech language pathology services delivered via TR (88, 89). The evidence suggests that TR will become an alternative service delivery mode for speech-language pathology.

Dysphagia is a symptom common to a wide range of medical conditions. Impairments of swallowing (dysphagia) may occur because of damage or dysfunction in the neurological control. Communication and swallowing disorders are highly prevalent in people affected by PD. Maintaining these functions over time becomes a challenge for PD patients and their caregivers. Although some emerging studies support the delivery of certain aspects of speech pathology practice via TR, the clinical assessment of dysphagia presents specific challenges for a TR model.

Only 2 preliminary reports by Sharma (90, 91) investigating patient satisfaction with regard to a specific TR-based treatment for dysphagia were identified. Sharma et al. (90) provided pilot information on the basic feasibility and validity of conducting dysphagia assessments via TR and examined the potential of TR for swallowing disorders. Ten simulated patients were assessed simultaneously face-to-face and by TR. The results were positive, with high levels of agreement observed between both groups on all parameters of interest (oromotor function and swallowing). Sharma et al. (91) evaluated patients' perception pre-treatment and their satisfaction after the TR treatment through pre- and post-session questionnaires in 40 patients with dysphagia (4 with PD). The 14 questions explored comfort with the use of TR, satisfaction, benefits of TR assessments, and the patient-preferred assessment modality. The results demonstrated that the use of TR in speech-language pathology, specifically in the assessment, and management of swallowing disorders, is a promising approach. It however requires insight from patients and clinicians in order to achieve optimal care. Patients had positive changes in their pre-assessment perceptions and had high levels of satisfaction with their experience. The data were positive and highlighted that patients are interested in and willing to receive services via TR. This provided preliminary evidence for the feasibility of remote dysphagia assessment. In the literature, only preliminary data collections in patients following laryngectomy were identified by Ward et al. (89). Recent evidence supports the feasibility, validity, and reliability of administering clinical dysphagia assessments via TR as opposed to face-to-face in several neurological conditions including PD (90–92).

PD can significantly alter the capacity to perform regular ADLs (e.g., self-care tasks) as well as work and leisure activities. Reduced independence in ADLs has been linked to a poorer QoL in people with PD (57). Some studies have evaluated how TR programs and, more generally, telemedicine can influence the QoL. Siegert et al. (68) investigated QoL as a primary outcome and found that providing remote specialist assistance directly into people's homes is feasible and improves healthcare, QoL, and social participation. These results show that a tablet-based training program can help to maintain long-term functional ability for PD. Dorsey et al. (75), in a telemedicine study, evaluated if the remote delivery of specialty care directly into people's homes can enhance access and improve the healthcare of individuals with chronic conditions, via virtual house calls. The authors showed that telemedicine can improve participants' QoL. QoL was also investigated in studies whose aim was to study the influence of TR on motor (30, 42, 56, 63) and non-motor symptoms (60, 67). The results indicate a short-term improvement in the QoL.

In addition, a few studies on motor (42, 63) and non-motor (58, 59, 64) symptoms have investigated patients' satisfaction with a TR program, reporting positive results. Although these results are positive, the number of studies analyzing patient satisfaction in association with TR is still too small to draw any meaningful conclusions. However, several studies exist that reported patient satisfaction following a telemedicine program. The study of Wilkinson et al. (61) is a dual-arm RCT focused on patient satisfaction as the primary endpoint (Patient Assessment of Communication of Telehealth questionnaire—PACT), as well as on clinical outcomes, patient travel burden, and health care utilization, using clinical video telehealth vs. usual inpatient care. Telemedicine studies, such as Venkataraman et al. (93), sought to characterize the recommendations and feedback of patients with PD obtained following a free consultation with a specialist via a virtual visit. The studies found high patient satisfaction with telehealth. Two studies by Sharma et al. (90) and Sharma et al. (91) had as primary endpoint the satisfaction of patients following participation in assessment of dysphagia conducted by TR. The studies showed patients' comfort and high satisfaction. Feasibility studies reported patient satisfaction following a telemedicine intervention.

High patient satisfaction was found in the studies by Antonini et al. (76) and Ferreira et al. (77), which also highlighted the ease of use of the treatment system. Barbour et al. (29) analyzed satisfaction not only of the patients, but also of family members, sub-specialists and nursing staff, reporting the same results.

TR systems for people with PD show some limitations. First, feasibility of TR is good in PD early stages and adult patients (not too old): patient's age, stage of disease and cognitive status are fundamental factors determining the success of a TR program. The majority of the studies cited in the present review found use of TR for assessment and treatment to be valid and reliable in the early stages of PD, but reported that it may not apply to individuals at an advanced stage of PD. To date, the findings regarding TR are similar to those for face-to-face treatment but there are no studies evaluating the long-term effectiveness of TR systems on motor symptoms. Secondly, long-term follow up are not available: while studies have demonstrated TR systems to be effective in the management of musculoskeletal disorders and neurological pathologies, in PD only preliminary studies are available, carried out on small numbers of patients. Finally, motor symptoms of PD (balance, gait, postural instability, UL dexterity) were evaluated with different scales in the different studies and different types of devices were applied (often not specifically set for PD patients), making it difficult to analyze accurately across the board the effectiveness of the TR systems. In contrast, for motor speech disorders a greater consensus about the methods, the scales and the devices used was evident.

The positive aspects of these preliminary studies are that at first TR has been proven to be suitable to treat both motor and non-motor symptoms in PD patients. Secondly, the majority of the studies reported here considered and measured the improvement in QoL and patient satisfaction after treatment.

Regarding future studies, based on the evidence to date, we would make two main recommendations. The standardization of the assessment scales used for PD motor symptoms is highly recommended. The scales most commonly used are: PDQ for QoL, UPDRS (embraces motor symptoms but does not embrace the pathology in all its complexity); BEST/mini-BESTest, TUG, 10-MWT, ABC for gait and balance and postural instability; 9HPT for UL dexterity. On the contrary, studies on non-motor symptoms (speech and voice) used a standardized system for research, which led to results that are more meaningful. Further studies on the efficacy of TR in the management of motor and non-motor symptoms of PD are also necessary.

This systematic review suggests that TR in PD patients is indicated in the early stages of disease and in particular in adult patients with preserved cognitive status.

As a general outcome, the indication is to carry out the TR treatment for 1 h/day, 3–4 days/weeks, for 4–12 weeks.

TR was demonstrated to be efficient in the improvement of specific outcomes since feedback or augmented reality embedded in this particular rehab-technique help the training of specific gestures stimulated by the achievement of a target, the correct execution of the movement and self-management.

This review highlighted that in order to improve the effectiveness of the treatment, a key factor is the use of devices that incorporate advanced technologies like visual/auditory feedback and augmented reality in order to specifically train a task. However, a comparative study amongst different technologies does not exist yet.

For the clinician it is of paramount importance to dispose of standardized and effective assessment tools (scales) and devices that use the same technology in order to normalize the treatment on a set of patients.

The patient must have an easy-to-use tool that provides feedback to guide him toward the achievement of the goal.

Through augmented feedback (VR, biofeedback, etc.) used in technologically assisted rehabilitation it was possible to whiteness an increase of the patient compliance and to train successfully tasks that are generally unaware like daily and sport gestures.