The effect of action observation training on gait and balance of patients with neurological and musculoskeletal disorders: A systematic review

Abstract

Background

Various treatment approaches are being applied for recovery of gait after different medical conditions. Action observation is a new motor learning approach, which is considered as a complementary training to the conventional rehabilitation programs such as occupational therapy for this purpose.

Objective

To find out which patients benefit more from action observation training.

Methods

Electronic databases, including Scopus, PubMed, Web of Science, Science Direct, and PEDro were searched. Prospective studies published in peer-reviewed journals with full text available in English, which investigated the effect of action observation on gait and balance of patients with neurologic or musculoskeletal disorders, were included. The methodological quality of the studies was assessed by the Downs and Black checklist, and the information was presented based on the PICO style.

Results

Nineteen studies recruiting post-orthopedic patients (4 studies), patients with stroke (11 studies), and Parkinson’s disease (4 studies) fulfilled the eligibility criteria. Quality scores ranged from 51.85% to 81.48%. Balance and walking ability were the most reported primary outcomes.

Conclusion

Patients in the chronic phase of stroke might benefit more from action observation training plus occupational therapy in different aspects of gait than orthopedic patients and those with Parkinson’s disease.

Keywords

Mirror neuron nervous system diseases postural balance walking

Introduction

The process of walking as a common and meanwhile complex activity of daily living depends upon the nervous system, musculoskeletal apparatus, and the cardiorespiratory system (Pirker and Katzenschlager, 2016). Locomotion, balance, and adaptability to the environment are the three main components of normal gait achieved through the proper balance amongst the involved neuronal systems. Although gait disorders are common in the old population, growing evidence suggests that such disorders are not only due to aging but also associated with common diseases of advanced age (Snijders et al., 2007). Neurological disorders, including stroke, Parkinson’s disease, and multiple sclerosis (MS) commonly result in gait disorders mostly characterized by reduced speed, and abnormalities in muscle tone and activation patterns in post-stroke patients (Chen et al., 2005; Pizzi et al., 2007) and restricted outside walking and physical activity of those with Parkinson disease and MS (Pelosin et al., 2010; LaRocca, 2011). In addition to neurological diseases, musculoskeletal disorders, including osteoarthritis and rheumatoid arthritis, are likewise considered among the leading causes of disability in older adults (Woolf and Pfleger, 2003). Degenerative changes in weight-bearing joints such as knees, and joint deformities like knee varus and valgus, are examples of musculoskeletal disorders affecting gait (Park et al., 2014b).

Besides, safety and stability of gait are generally considered as key indicators of functional independence in the elderly (Krebs et al., 1998). The CNS-related impairment in the coordination of motor outputs is one of the main reasons of gait instability in patients suffering from neurological conditions (Hollman et al., 2007).

Assisting movements, muscle strengthening, and modifying current gait strategies are common gait and balance rehabilitation techniques to help patients maintain their motor coordination (Patel, 2017). Hence, novel physiologic-based treatment programs, focused on rebuilding the neural circuits which underlie the functional impairments, have emerged recently (Small et al., 2013). Along these lines, action observation (AO) is a new behavioral intervention that may potentially affect neural circuit reorganization (Small et al., 2012).

Novel findings in neuroscience suggest that motor actions can be influenced by observing others doing the same action (Mattar and Gribble, 2005). This idea is based on the mirror neuron system (MNS), a group of neurons discovered at first in the premotor cortex of the monkey, that activated both during performing a hand motor task by a monkey and observing a similar action done by others (Ferrari et al., 2003). The presence of a similar observation/execution matching system was first discovered in humans in a magnetic stimulation study (Fadiga et al., 1995). This system is also considered as an essential part of imitation-based learning, a cognitive function involved in learning a new motor skill (Buccino et al., 2004).

The contribution of MNS in memory formation of an observed action (Stefan et al., 2005), improvement of motor performance in terms of force production (Porro et al., 2007), and stronger activation of motor representation of acquired motor skills of observers (Merino et al., 2005) suggest that this system potentially retain an application in motor rehabilitation. Furthermore, the findings of MNS suggest that this system is highly influenced by several principles emphasized in occupational therapy. The MNS as a multimodal, goal-directed, context-dependent neural system with an experience-dependent modulation capacity may be highly applicable for motor rehabilitation in occupational therapy.

Over the past few years, action observation therapy (AOT), in terms of observing exercises before execution or simultaneously, has been used in motor rehabilitation of both upper and lower limb dysfunctions. Supportive evidence in favor of the value of AOT as an add-on treatment to the conventional rehabilitation programs, including occupational therapy for upper limb function for patients with different neurological deficits such as stroke and cerebral palsy has emerged from earlier studies (Franceschini et al., 2012; Buccino et al., 2012). The efficacy of AOT for the restoration of gait and balance has also been evaluated in different conditions, including post-surgical orthopedic patients (Bellelli et al., 2010), Parkinsonians (Pelosin et al., 2010), and post-stroke patients (Bang et al., 2013).

Regarding the various underlying pathologies in different populations recruited in these studies, this question arises that which patients may benefit the most from AOT in the improvement of gait and walking abilities.

The present systematic review used a qualitative approach to synthesize the available evidence to assess the added value of AOT as a new physiology-based rehabilitative technique on different aspects of gait and balance recovery of neurologic and orthopedic patients with gait disorders.

Materials and Methods

Inclusion and exclusion criteria

To frame the eligibility criteria and to define the search terms, the PICO (P: Population; I: Intervention; C: Comparison; O: Outcome;) framework was used. P was defined as “neurologic or orthopedic patients with gait disorder”; I as “action observation training”; C as “physical therapy alone or no treatment”; O as “spatiotemporal gait parameters, walking ability, and balance.”

Search strategy

To isolate relevant evidence, electronic databases, including PubMed, Scopus, Web of Science, Science Direct, PEDro, and DOAJ were comprehensively searched in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines from their origin to 2018. A combination of keywords related to inclusion criteria of the study were used. For example, the search in PubMed was done as follows:

(Action observation [TIAB] OR Mirror neurons [TIAB]) AND (Walking [TIAB] OR Gait [TIAB] OR Balance [TIAB] OR Lower limbs).

To identify other eligible studies, a search through the list of references of included studies was carried out.

After the search, titles and abstracts of all papers were reviewed by two reviewers (FSH and MK) for identifying relevant studies. The final list of studies included those that met the following criteria:

1. Studies that were published in peer-reviewed journals with available full-text in English.

2. Studies that evaluated the effects of multi-session action observation training in a group of patients before and after the intervention and randomized clinical trials.

3. Case studies and case reports were excluded.

4. Studies in which subjects were patients with neurological diseases or musculoskeletal disorders.

5. Studies in which AOT was the intervention of interest.

6. Studies that included spatiotemporal, kinetic, and kinematic parameters, walking abilities (10-Meter Walk Test, 6-Min Walk Test, Figure-of-8 Test), as well as balance and postural control as dependent variables.

When there was no sufficient data in the abstract, full-text was considered for inclusion.

Quality assessment and determination of the risk of bias: The Downs and Black quality checklist was used to assess the methodological quality of the included studies. This checklist is for evaluating the quality of both randomized and non-randomized studies, which has high internal consistency and high test-retest reliability. It also has good inter-rater reliability; (KR-20: 0.54), (r = 0.88), and (r = 0.75), respectively (Down and Black, 1998). Two independent reviewers (FSH and MK) performed the quality assessment of all included studies. The Pearson correlation test was used to check the correlation between the scores, and a proper concurrence among the reviewers was found. Disagreements were discussed and partly resolved during three consensus meetings. The final quality scores were presented in percentage and were classified as high (>75%), moderate (60%–75%), or low (<60%).

Data extraction and analysis

Two reviewers (FSH and MK) extracted data from the included studies independently for descriptive analyses. The extracted data contained the description of study participants (mean age, type of disease, and duration of illness), type and duration of interventions, assessments, and outcomes. The focus of this review was only on the description and qualitative synthesis of the included studies, and meta-analysis was not performed due to the methodological differences in terms of tests, tasks, and outcome measures.

Results

A total of 1040 studies were initially selected through electronic search across databases. Study titles and abstracts were reviewed for eligibility, and duplicates were removed. This search resulted in 26 papers from which three other studies were excluded since full texts were unavailable (Buccino et al., 2011; Rezzan et al., 2015; Sale et al., 2018), and one was excluded because it was a case report (Chatterton et al., 2008). Three other studies were also excluded because their full text was not in English (Ghanjal et al., 2015; Ghanjal et al., 2014; Takeshi et al., 2015). Eventually, 19 eligible reports were included in this review (Figure 1). From the 19 reviewed papers, 11 had recruited stroke patients with hemiparetic gait, four were on post-operatic patients who had undergone hip or knee arthroplasty, and four dealt with patients with Parkinson’s disease (Tables 1–3). While only one investigation included a single group of patients (Santamato et al., 2015), others included at least one experimental and one control group.

Figure 1.

Flowchart of search strategy and study selection.

Table 1.

A summary of the studies on stroke patients.

Reference	Method	Assessments	Outcome
(Kim and Kim, 2012)	Subjects: 30 stroke patients (hemiparesis; infarction/Hemorrhage), 15 in each group Side of stroke: Right/Left: 8/7 each group Mean age (year):AOG: 64.1 ± 8.3/CG: 65.5 ± 7.7 Time after onset (month): AOG: 4.6 ± 1.3/CG: 4.1 ± 1.0 Type of intervention: Observation of walking straight with long steps in slow and fast motion Duration of intervention: 30min PT + 10min AO; number of sessions not mentioned	Spatiotemporal gait parameters, GAITRite (affected side step and stride length and single support time, double support time, gait velocity, cadence)	Significant improvement of the step length, stride length, single support time, double support time, gait velocity, and cadence in AOG.
(Bang et al., 2013)	Subjects: 30 stroke patients (hemiparesis; infarction/Hemorrhage), 15 in each group Side of stroke: Right/Left: AOG 8/7, CG: 6/9 Mean age (year): AOG: 64.1±6.35/CG: 58.9±7.03 Time after onset(month): AOG: 14.1±3.78/CG: 12.6±3.86 Type of intervention: Treadmill training after watching a treadmill video Duration of treatment: 40 min per day (9 min watch/30min exercise)/5 times per week for 4 weeks	TUG test,10-MWT, The 6-MWT, The maximal flexed knee angle	Significant differences between the groups for TUG test, 10-MWT, 6MWT and KASW
(Kim and Lee, 2013)	Subjects: 27 stroke patients (hemiparesis; infarction/Hemorrhage), 9 AOT/9 PT/9 MIT Side of stroke: Right/Left: AOT:6/3, PT: 3/6, MIT:5/4 Mean age (year): AOPT: 55.3±12.1, PT:59.8±8.9, MIT:54.8±8.8 Time after onset(month): AOT: 8.3±3.3/PT: 8.5±3.6/MIT: 7.3±0.7 Type of intervention: Observation of videos showing rotation in the sitting position/sit to stand and stand to sit/weight shift to the front and back, left and right, and weight shift involved lifting a foot on the block while standing. Gait level surface and step over an obstacle for improvement of gait ability, pelvic tilting, trunk flexion, and extension Duration of intervention: 30-min (20min watch/10min PT)/5 times per week + neurodevelopmental therapy for 30 min, twice a day, 5 days per week for 4 weeks	Spatiotemporal gait parameters using a GAITRite TUG test, the functional reaching test, the walking ability questionnaire, the functional ambulation category	A significant difference between the AOT group and the physical training group in TUG/gait speed/cadence/single-limb support. No significant differences among the three groups in walking ability questionnaire and functional ambulation category

(Park and Kang, 2013)	Subjects: 20 stroke patients (hemiplegia), 10 AOG/10 CG Right/Left: 4/6 in each group Mean age(year): Not exactly mentioned Time after onset: More than 6 month Type of intervention: Experimental group: Functional electrical stimulation (FES)treatment + video on gait showing walking on flat ground, slopes, and stairs control group: Just FES treatment Duration of intervention: 15 min AO, 5 times per week for 6 weeks	Weight bearing and stability	Significant changes in front and rear foot weight bearing, left and right foot weight bearing, stability index and gait velocity in experimental group. A significant difference in anterior and posterior weight bearing in control group. Between-group differences in RL, SI, gait velocity
(Park et al., 2014a)	Subjects: 21 stroke patients (hemiparesis; infarction/hemorrhage), AOG: 11/CG:10 Side of stroke: Right/Left: AOG: 4/7; CG: 4/6 Mean age(year): AOG: 55.91±9.10/CG: 54.80±12.22 Time after onset (month): AOG: 21.09±16.66/CG: 25.60±19.67 Type of intervention: Observation of walking tasks that frequently experienced in pre-morbid life, weight shifting to the affected side, walking on straight and curved paths, even and uneven surfaces, crossing obstacles, and walking with functional tasks (normal and slow speeds) Duration of intervention: 10 min AO + 20 min PT/3 times per week for 4 weeks	Symmetry scores (spatiotemporal parameters of GAITRite) walking ability (10 MW/F8W/DGI)	10MWT, F8WT, and DGI were more favorable in the experimental group than the control group, p<0.05. No significant differences for all variables between pre- and post-test in the control group. No significant differences in gait symmetry scores
(Park and Hwangbo, 2015)	Subjects: 40 stroke patients (hemiplegia), 20 in each group Side of stroke: Right/Left: AOGT: 9/11; GGT: 11/9 Mean age (year): AOGT: 51.15±14.81/GGT: 48.65±12.81 Time after onset (month): AOGT: 14.8±6.1/GGT: 13.4±8.2 Type of intervention: Observation of three-minute videos each for walking on a flat land, on a slope, and steps Duration of intervention: 30 min of general physical therapy +10min watching video+20 min walking exercise/5times per week for 8 weeks	The balance ability measuring and training system (sway area, sway speed, and limit of stability in a static standing position) TUG, 10MWT	Significant differences in the sway speed and total length of the limit of stability and TUG test and the 10MWT between the two groups after the experiment. No significant differences in the sway area between the groups

(Park et al., 2017)	subjects: 25 stroke patients (Infarction/hemorrhage), AOG: 12; CG: 13 Side of stroke: Right/Left: AOG: 7/5; CG: 4/9 Mean age (year): AO: 57.33 ± 6.89/CG: 55.08 ± 8.12 Time after onset(month): AOG: 33.58 ± 24.67 CG: 21.15 ± 12.44 Type of intervention: Observation of walking on even and uneven ground, in a complex and unpredictable community environment, in a parking lot and shopping center at normal speed and half times the normal with sound Duration of intervention: functional training 30 min per day/5 times per week and observation of 30-min video clips, three times per week for 4 weeks	Spatiotemporal gait measures: GAITRite (the cycle time (s), stride length (m), single support (%GC), double support (%GC), and velocity (cm/s)) 10MWT, Community walk test, Activities-specific balance confidence scale	Significant improvements in the 10-MWT, community walk test, and activities-specific balance confidence score after intervention in the experimental group. In CG only in 10MWT. Significant differences between the two groups in stride length, single support, and velocity
(Song and Lee, 2016)	Subjects: 16 stroke patients (hemiparesis),8 in each group Side of stroke: Right/Left: TAG:3/5, TLG: 4/4 Mean age (years): TAG: 56.25±8.2/TLG: 59.75±10.06 Time after onset(month): TAG: 32.25±26.92 TLG: 41.12±29.62 Type of intervention: Observation of treadmill walking, indoor, outdoor, ascent, downhill walking, avoid obstacles, skipping obstacle, uneven surface in 0.8 m/s walking speed and 1.1–1.5 m/s speed Duration of intervention: AO + treadmill simultaneously 20 min, once a day, 3 times per week for 4 weeks	10 MWT, 6 MWT, DGI	Significant difference was found post- to pre-test in both groups. A significant improvement was observed in TAG compared with TLG just in 10MWT and DGI. No significant difference in 6MWT between groups

(Lee et al., 2017)	Subjects: 35 stroke patients,AOTA:12, MTA:11, AOT:12 Mean age (years): AOTA: 62.8±7.4/MTA: 57.27±5.7 AOT: 59.8±6.7 Time after onset(month): more than six months Type of intervention: Observation of dorsiflexor training composed of knee joint extensor and dorsiflexor training, knee joint flexor and dorsiflexor training, hip and knee joint flexor and dorsiflexor training. Duration of treatment: AOTA: 15 min observation +15 min activity/3 times per week for 6 weeks/AOT: 30 min observation + general PT, 30 min a day twice a week	Biodex Balance System, postural stability and fall risk, Modified Functional Ambulation Profile FOR task-specific gait ability	No significant difference between groups in all balance variables just more improvement in AOTA than pre-test. No significance between groups in gait parameters just different post than pretest in AOTA and MTA.
(Motaqhey et al., 2015)	Subjects: 24 stroke patients, 12 Male, 12 Female Side of stroke: Right/Left: Women: AOG: 1/5; CG.:4/2 Men AOG:2/4; CG.:1/5 Mean age (years): Women AOG: 54.8±6.2/CG.:54.3±7.3 Men AOG: 54.8±7.4/CG.: 54.7±7.6 Time after onset(month): Women AOG:5.5±1/CG: 5.8±0.7 Men AOG: 5±0.9/CG: 4.7±0.8 Type of intervention: Observation of video clips of movements routinely used for physiotherapy of hemiparetic patients Duration of treatment: AOG: 15 min watching+ 45 min routine PT, CG: just 45 min PT/3 times per week for 4 weeks	TUG, 6-MWT, Barthel index, BBS	Improvement of TUG, SMW, BI, BBS in all groups highest in observation groups with more improvements in female watching than other groups
(Kim and Lee, 2015)	Subjects: 12 stroke patients (infarction/hemorrhage), SPOG=5/AOG = 6 Side of stroke: Right/Left: AOG:4/2; SPOG: 5/1 Mean age (years): AOG: 56.33±4.45/SPOG: 60.33±8.09 Time after onset (month): More than 6 months Type of intervention: Observation of videotapes composed of the STS and the TUG movement by AOT group and static scenery pictures by SPO group Duration of intervention: 5 min watch +25 min practice each session, 3 times per week for 4 weeks	Sit to Stand test, DGI: Evaluating balance and ambulation, TUG test	Significant differences in the TUG capability of the AOT group and the SPO group. No significant differences between groups in the STS test and DGI

AOG: action observation group; CG: control group; PT: physical therapy; TUG: timed up and go; 10-MWT: 10-meter walk test; 6-MWT: 6-minute walk test; KASW: knee angle in swing phase during walking; AOT: action observation training; MIT: motor imagery training.

RL: right left; SI: stability index; F8WT: figure-of-8 walk test; DGI: Dynamic Gait Index; AOGT: action observational gait training; GGT: general gait training.

BT4: Balance trainer 4; BBS: berg balance scale; GAOT; group action observation training; IAOT: individual action observation training; GC: gait cycle; AG: Treadmill training applied simultaneously with action observation; TLG: Treadmill training applied simultaneously with landscape observation.

AOTA: action observation therapy with activity; MTA: mirror therapy with activity; AOT: action observation therapy; BI: Barthel index; SPOG: scenery picture observation; STS: sit to stand.

Table 2.

A summary of the studies on orthopedic patients.

Reference	Method	Assessments	Outcome
(Bellelli et al., 2010)	Subjects: 60 patients with a hip fracture, or hip or knee replacement, 30 in each group Mean age (years): AOG. 71.9±8.4/CG. 71.8±6.9 Time from surgery (days): AOG: 7.7±2.2/CG: 7.3±2.4 Type of intervention: Observation of daily actions with the leg or trunk for experimental group and scenes with no motor content for controls Duration of intervention: Conventional post-operatic rehabilitation for 1 hour a day/6 days per week for 3 weeks+ 24 min observation leg and trunk actions or natural scenery	FIM score, the motor FIM sub-score (the mobility and locomotion items), the Tinetti scale score, dependence on walking aids for gait	FIM absolute functional efficiency, FIM score minus relative functional gain, and FIM motor subscores were significantly greater in patients of the case groups than in patients of the control groups. A significant improvement in the case group was observed for both mobility and locomotion
(Park et al., 2014b)	Subjects: 18 patients with knee arthroplasty, 9 in each group Mean age (years): AOG: 72.67 ±12.25/CG: 70.56 ±10.98 Time after surgery: Not mentioned Type of intervention: Observation of raising legs in a supine position, flexing and extending the ankles, sitting without flexing the knees, sitting and standing without flexing the waist, walking with a walker, going upstairs and downstairs, flexing and extending the knees, and standing on one leg with the hands against the wall Duration of intervention: AOG: 10 min observation +30 min PT, CG: 30 min PT once a day/3 days per week for 3 weeks	Evaluation of pain, stiffness, and one activity of daily living using WOMAC, TUG	No significant differences in the TUG result between the two groups. Significant differences of WOMAC items including stiffness and function between groups
(Villafane et al., 2017)	Subjects: 31 patients with total knee replacement, AOG:14, CG:17 Side: Right/Left; AOG: 6/8, CG:14/3 Mean age (years): AOG: 70.4±7.5/CG: 70.1±7.7 Time after surgery (days): AOG: 6.8±1.4/CG: 8.3±1.5 Type of intervention: Observation of exercises included in conventional physiotherapy consisted of mobilization, transfer practice, and exercises for isometric and isotonic strengthening of muscles. Duration of intervention: Observation of a video of the exercises being performed by AOG and a video of nature scenes by CG + 30 min conventional PT in addition to 20 min Continuous Passive Motion Device twice a day/5 days per week for 2 weeks	Active range of motion, passive range of motion, Barthel index, Cumulative Illness Rating Scale, Short Form-36 Health Survey, Tinetti scale, and Lequesne	No group interaction for pain intensity. Higher flexion and extension range of motion in the experimental group. No group interaction for passive range of motion, Functional status and gait assessed by Barthel index; Tinetti scale, and Lequesne index, however, between-groups effect sizes were moderate to great at post-treatment period

(Villafane et al., 2016)

Subjects: 24 patients with primary total hip arthroplasty, AOG: 9, CG: 15
Mean age (years): AOG: 75.4±5.3/CG: 64.9±7.4
Side: Right/Left: AOG:7/2; CG: 15/0
Time after surgery(days): AOG: 7.1±2/CG: 7.7±2.1
Type of intervention: Observation of active mobilization of the lower limb in lying, sitting and standing positions, and some daily living activities, such as the transfer from lying to sitting and from sitting to standing
Duration of intervention: AOG: 15 min exercise +15 min observation/CG: 15 min exercise + written information, twice a day/5 days per week for 2 weeks

The flexion and abduction active range of motion and passive range of motion, the Barthel Index, the Cumulative Illness Rating Scale, the Short Form (SF-36) Health Survey, the Tinetti scale, the Lequesne Index

Between-group effect sizes were small for pain intensity, flexion and abduction active and passive range of motion, Functional Status including The Tinetti Scale and the Lequesne Index outcomes, for quality of life. Between-group effect sizes were greater in the post-treatment period

FIM: functional independence measure; WOMAC: western Ontario and mc master universities osteoarthritis index.

Table 3.

A summary of the studies on Parkinson’s disease.

Reference	Method	Assessments	Outcome
(Pelosin et al., 2010)	Subjects: 18 patient with Parkinson’s disease, 9 in each group Mean age (years): AOG: 68.8±4.1/CG: 70.2±6.8 Disease duration (years): AOG: 11.6±4.9/CG: 9.5±3.7 Type of intervention: Observation of strategies useful in circumventing FOG episodes by AOG and landscape videos by CG Duration of intervention: A 60-min physical therapy training (24 min observation +36 min PT)/3 sessions per week for 4 weeks	FOG Questionnaire and FOG diary, TUG, 10MWT, Tinetti scale part 1, BBS, Tinetti scale part 2	Significant reduction of the number of freezing episodes only in the Action group, particularly in the start walking and turning conditions. A significant improvement of walking (TUG, 10M-WT, Tinetti part I) and balance (BBS, Tinetti part II) tests, as well as the PDQ-39 in both groups without between-group differences
(Santamato et al., 2015)	Subjects: 14 patient with idiopathic Parkinson’s disease Mean age (years): 68.9 ± 4.7 More affected body part: Right/Left: 10/5 Time from diagnosis(years): 3.5±1.9 Type of intervention: Observation of unsupported sitting on the table with upper limbs abducted to 90°, picking something off the floor, walking in straight and not straight path, unsupported standing with one foot, walking sideways, overcoming an obstacle, getting round obstacles Duration of intervention: 40 min each session: 20 min observation +20 min PT, 3 times per week for 8 weeks	BBS, The Activities-Specific Balance, Confidence Scale UPDRS III, 10MWT, TUG	No statistically significant differences were observed in all primary and secondary outcome measures
(Agosta and Sarasso, 2016)	Subjects: 25 patients with idiopathic Parkinson’s disease, AOG: 12/CG: 13 Mean age (years): AOG: 69.0±8.0/CG: 64.0±7.0 Type of intervention: Observation of videos of strategies useful in circumventing FOG episodes + auditory cues by AOG and videoclips containing sequences of static pictures of landscapes by CG Duration of intervention: A 60-min physical therapy training (24 min of observation and 36 of imitation)/3 sessions per week for 4 weeks	(UPDRS) III, HY scale, FOG-Q and UPDRS II FOG Score, quality of life with the 39-item PD questionnaire, BBS, the 10 MWT	At week 4, the AOT group showed reduced FOG severity and improved UPDRS III ON, and PDQ39, BBS and 10 M-WT scores, while the “Landscape” group showed a reduced FOG severity, decreased PDQ39 score, improved walking abilities and a trend toward an improved BBS score. At week 8 positive effects on UPDRS III ON, PDQ39 and balance scores were maintained in AOT group only

(Jaywant et al., 2016)

Subjects: 23 patients with idiopathic Parkinson’s disease, AOG: 13, CG: 10
Mean age (years): AOG:63.7±6.2/CG: 65.8±8.7
Type of intervention: Observation of videos of actors with and without PD, walking in a hallway from lateral and anterior, posterior views by AOG and videos of landscapes with moving water with no biological motion by CG
Duration of intervention: One week of AO of the same videos daily

Laboratory-Based and Home Walking Assessment, Spatiotemporal gait Parameters, walking speed, stride length, and stride frequency, leg swing time and gait asymmetry and Self-Report Questionnaires

Increased self-perceived mobility (small effect) following the training for AOG but the lack of objective change.
No main or interaction effects for walking speed, mean stride length and mean stride frequency, for percent swing time or gait asymmetry, walking periods per hour, mean duration of each walking period

FOG: freezing of gait; PDQ-39: the 39-item PD questionnaire; UPDRS III: unified Parkinson’s disease rating scale.

Risk of bias

The quality of all included studies was assessed by the Downs and Black quality checklist. Pearson correlation test was used to find the agreement between two reviewers’ assessments (FSH and MK). The results demonstrated a proper agreement (r= 0.823).

The range of total quality scores was from 51.85% to 81.48%. (Mean ± standard deviation, 66.6±8.92%). According to the adopted categories of 19 assessed papers, three studies were of high quality (scored 77.77% to 81.48%) (Bang et al., 2013; Motaqhey et al., 2015; Bellelli et al., 2010), 11 were of moderate quality (scored 62.96% to 74.07%) (Jaywant et al., 2016; Agosta and Sarasso, 2016; Park et al., 2014a, 2014b, 2017; Kim and Lee, 2015; Villafane et al., 2016, 2017; Pelosin et al., 2010; Song and Lee, 2016; Kim et al., 2014), and five were of low quality with scores ranging from 51.85 to 59.25% (Kim and Kim, 2012; Lee et al., 2017; Park and Kang, 2013; Santamato et al., 2015).

The reporting sub-score ranged from 60 to 90%. All included studies but one (Pelosin et al., 2010) failed to report adverse effects of the intervention of interest. The other item not reported in 10 studies was the distribution of principal confounders (Park and Kang, 2013; Jaywant et al., 2016; Agosta and Sarasso, 2016; Santamato et al., 2015; Pelosin et al., 2010; Villafane et al., 2016; Park et al., 2014a, 2014b; Lee et al., 2017; Kim and Kim, 2012). Characteristics of patients, who were lost to follow-up, were not mentioned clearly in seven studies (Kim and Lee, 2015; Santamato et al., 2015; Villafane et al., 2016; Park et al., 2014b; Lee et al., 2017; Park and Hwangbo, 2015). Furthermore, seven studies did not report actual probability value (Park and Kang, 2013; Jaywant et al., 2016; Pelosin et al., 2010; Lee et al., 2017; Park and Hwangbo, 2015; Park et al., 2014a; Kim and Kim, 2012). Also, some studies neither reported the characteristic of patients nor the hypothesis and aim/objective of the study (Park and Kang, 2013; Kim and Kim, 2012). Regarding the external validity, the source of populations was not identified in any of the studies, and also there was no explanation in the studies to show whether the included participants were representative of the entire study population. Meanwhile, the type of intervention used in all studies but one (Jaywant et al., 2016) could be achieved in the source population. The internal validity-bias sub-score ranged from 57% to 100%. Only one study reported all items clearly (Bang et al., 2013). In addition, just five studies blinded their subjects to the intervention performed (Kim and Lee, 2015; Jaywant et al., 2016; Motaqhey et al., 2015; Park et al., 2014a; Bang et al., 2013). In seven studies, the main outcomes were measured by blinded examiners (Bang et al., 2013; Park et al., 2017; Bellelli et al., 2010; Villafane et al., 2016, 2017; Pelosin et al., 2010; Agosta and Sarasso, 2016). Besides, in four studies, the validity and reliability of the main outcome measures were not clearly described (Kim and Kim, 2012; Park and Kang, 2013; Kim and Lee, 2015; Lee et al., 2017). The compliance with the treatment was unclear in one study (Park et al., 2014a), and one study used potentially inappropriate statistical analysis (Santamato et al., 2015).

The internal validity for confounding variables scores varied from 33.33% to 83.33%. The source of population for both control and experimental groups was the same in all studies. In seven studies, it was clear that patients in both groups were recruited over the same time (Motaqhey et al., 2015; Lee et al., 2017; Bellelli et al., 2010; Villafane et al., 2016, 2017; Santamato et al., 2015; Jaywant et al., 2016).

The power sub-scale in the Down and Black checklist is concerned with the power of the study to detect a clinically important effect where the p-value for a difference due to chance is less than 5%. To answer this item, papers are checked for sample size calculation, and if the power or sample size calculation details have been reported power item in Down and Black is answered 1; otherwise, it is scored 0, and the percentage would be reported as 100% and 0%, respectively. Only two studies had mentioned clearly about the calculated sample size (Jaywant et al., 2016; Motaqhey et al., 2015). Random allocation, assignment concealment, and confounding adjustment were other items that differed across studies. The quality scores of the included studies are outlined in Table 4.

Table 4.

Methodological quality scores of the included studies.

Author and year	Reporting, %	External validity, %	Internal validity, %		Power, %	Overall quality score, %
Author and year	Reporting, %	External validity, %	Bias confounding, %		Power, %	Overall quality score, %
(Kim and Kim, 2012)	60	33.33	57	66.66%	0	55.5
(Bang et al., 2013)	90	33.33	100	83.33%	0	81.48
(Kim and Lee, 2013)	90	33.33	71.42	50%	0	66.66
(Park and Kang, 2013)	60	33.33	57	66.66%	0	55.5
(Park et al., 2014a)	70	33.33	71.42	66.66%	0	62.96
(Park and Hwangbo, 2015)	70	33.33	71.42	50%	0	59.25
(Park et al., 2017)	90	33.33	85.71	66.66%	0	74.7
(Song and Lee, 2016)	90	33.33	71.42	66.66%	0	70.37
(Lee et al., 2017)	60	33.33	57	50%	0	51.85
(Motaqhey et al., 2015)	90	33.33	85.71	83.33%	100	81.48
(Kim and Lee, 2015)	80	33.33	71.42	33.33%	0	62.96
(Bellelli et al., 2010)	90	33.33	85.71	83.33%	0	77.77
(Park et al., 2014b)	80	33.33	71.42	66.66%	0	66.66
(Villafane et al., 2017)	90	33.33	85.71	66.66%	0	74.07
(Villafane et al., 2016)	70	33.33	85.71	50%	0	62.96
(Pelosin et al., 2010)	80	33.33	85.71	66.66%	0	70.37
(Santamato et al., 2015)	70	33.33	57	66.66%	0	59.25
(Agosta and Sarasso, 2016)	80	33.33	85.71	66.66%	0	70.37
(Jaywant et al., 2016)	80	0	85.71	83.33%	100	74.07

Outcome details

Spatiotemporal gait parameters

Of 19 studies included in this review, five studies evaluated the spatiotemporal parameters following AOT to quantify the variance of gait (Kim and Kim, 2012; Kim and Lee, 2013; Park et al., 2014a, 2017; Jaywant et al., 2016). Three studies showed improvement of single-limb support and gait velocity in AO group compared with physical therapy alone in patients with stroke (Kim and Kim, 2012; Kim and Lee, 2013; Park et al., 2017) (Table 5), and two studies reported improvement of stride length following AO (Kim and Kim, 2012; Park et al., 2017) (Table 5).

Table 5.

Summary data (mean± SD/median (IR), sample size) and significance of within and between-group differences of average scores of spatiotemporal gait parameters following action observation training.

	Action observation group			Control group
	Pre-intervention	Post-intervention		Pre-intervention	Post-intervention
Spatiotemporal gait parameters, patients with stroke	Mean ± SD/Median (IR)	Mean ± SD/Median (IR)	N	Mean ± SD/Median (IR)	Mean ± SD/Median (IR)	N
(Kim and Kim, 2012)
Step length of affected side (cm)	34.52 ± 11.69	41.55 ± 9.12^*†	15	34.29 ± 12.73	34.37 ± 8.61	15
Stride length of affected side (cm)	75.14 ± 11.67	89.23 ± 8.45^*†	15	72.41 ± 14.76	73.74 ± 14.42	15
Single leg support time of affected side (%)	31.65 ±9.77	38.52 ± 3.75^*†	15	29.85 ± 10.45	30.65 ± 9.45	15
Double leg support time (%)	36.78 ± 10.61	27.21 ± 6.43^*†	15	37.11 ± 10.65	33.21 ± 6.18	15
Gait velocity (m/s)	0.53 ± 0.17	0.73 ± 0.2^*†	15	0.48 ± 0.26	0.53 ± 0.15	15
Cadence (step/min)	73.49 ± 13.78	90.36 ± 13.8^*†	15	66.27 ± 16.41	75.49 ± 15.72	15
(Kim and Lee, 2013)
Step length of affected side (cm)	40 ± 6.96	46.76 ± 7.89^*	9	39.45 ± 8.49	42.92 ± 7.41^*	9
Stride length of affected side (cm)	79.62 ± 14.02	91.4 ± 15.77^*	9	78.45 ± 16.68	84.12 ± 16.04	9
Single leg support time of affected side (%)	27.07 ± 5.28	30.2 ± 4.5^*†	9	26.60 ± 4.59	27.43 ± 4.79	9
Double leg support time (%)	37.1 ± 5.45	36.29 ± 5.29^*		38.32 ± 4.01	38.06 ± 4.02	9
Gait velocity (m/s)	59.4 ± 16.64	78.67 ± 24.64^*	9	55.43 ± 16.64	62.88 ± 23.11^*	9
Cadence (step/min)	75.49 ± 22.25	97.02 ± 29.7^*†	9	72.65 ± 9.86	80.41 ± 15.87^*	9
(Park et al., 2017)
Stride length of affected side (cm)	57.4±7.85	76.3±17.27^*†	12	68.18 ± 24.59	71.34 ± 22.32	13
Single leg support time of affected side (%)	17.76 ± 5.79	23.6 ± 7.13^*†	12	23.12 ± 7.82	23.36 ± 7.72	13
Double leg support time (%)	52.83 ± 10.53	44.6 ± 10.5^*	12	47.82 ± 16.14	45.70 ± 12.99	13
Gait velocity (m/s)	31.80 ± 12.16	47.46 ± 19.5^*†	12	40.50 ± 23.14	43.46 ± 22.51	13
(Park et al., 2014a)
Swing phase asymmetry	0.98 (0.69–1.50)	0.60 (0.37–0.86)	11	0.61 (0.32–1.24)	0.90 (0.23–1.40)	10
Stance phase asymmetry	0.19 (0.12–0.22)	0.16 (0.07–0.19)^*	11	0.15 (0.10–0.21)	0.14 (0.09–0.21)	10
Stride length asymmetry	0.32 (0.12–0.38)	0.16 (0.09–0.48)	11	0.17 (0.13–0.42)	0.26 (0.09–0.41)	10
Spatiotemporal gait parameters, patients with Parkinson’s disease
(Jaywant et al., 2016)
Stride length (m)	1.34 ± 0.18	1.35 ± 0.21	13	1.3 ± 0.15	1.34 ± 0.12	10
Gait velocity (m/s)	1.19 ± 0.15	1.19 ± 0.15	13	1.13 ± 0.14	1.18 ± 0.08	10
Gait asymmetry	0.03 ± 0.02	0.03 ± 0.02	13	0.02 ± 0.01	0.02 ± 0.01	10

SD: Standard deviation; IR: Interquartile range; N= Number in each group.

* Significant within-group difference, p<0.05; †Significant between-group difference, p<0.05.

Improvement of cadence and double support were also reported by Kim and Kim (2012) and Kim and Lee (2013) (Table 5). There was just one study on Parkinson’s disease which reported no main or interaction effects for any of the spatiotemporal gait parameters following home-based AO in these patients (Jaywant et al., 2016) (Table 5). Park et al. (2014a) evaluated the gait symmetry scores, including swing phase, stance phase, and stride length in their study on chronic stroke patients, but their results showed no significant difference between the AO and control groups (Table 5).

Active and passive range of motion

AOT in groups of patients after hip and knee arthroplasty had no significant effects on the active and passive range of motion of hip and knee compared with conventional physical training alone. However, significant within-group differences were found in both groups (Villafane et al., 2016, 2017) (Table 6).

Table 6.

Summary data (mean± SD/median (IR), sample size) and significance of within and between-group differences of average scores of joints range of motion (ROM) and functional status following action observation training.

	Action observation group
	Pre-intervention	Post-intervention	N	Pre-intervention	Post-intervention	N
Active and passive ROM/functional status, orthopedic patients
(Villafane et al., 2016)
Active ROM of hip flexion	53.8 ± 13.0	83.8 ± 14.1^*	8	58.1 ± 13.8	80.8 ± 9.8^*	15
Active ROM of hip abduction	16.3 ± 15.5	32.5 ± 10.0^*	8	15.4 ± 12.0	27.1 ± 11.0^*	15
Passive ROM of hip flexion	65.6 ± 18.4	94.4 ± 11.2^*	8	72.9 ± 9.7	87.7 ± 4.4^*	15
Passive ROM of hip abduction	30.6 ± 5.6	38.8 ± 6.4^*	8	24.2 ± 6.7	32.5 ± 8.9^*	15
Barthel Index	58.5 ± 12.9	90 ± 6.8^*	8	62.9 ± 13.7	86.6 ± 6.9^*	15
Tinetti scale	15.9 ± 5.6	23.8 ± 2.7^*	8	16.8 ± 4.0	21.2 ± 3.9^*	15
Lequesne Index	19.9 ± 2.6	10.8 ± 3.1^*	8	18.8 ± 2.5	13.6 ± 2.9^*	15
(Villafane et al., 2017)
Active ROM of knee flexion	60.0 ± 18.8	95.7 ± 8.6^*	14	52.9 ± 16.4	80.1 ± 13.7^*	17
Active ROM of knee extension	−5.7 ± 3.5	−0.4 ± 1.2^*	14	−6.4 ± 4.1	−3.7 ± 3.5^*	17
Passive ROM of knee flexion	77.1 ± 17.8	98.8 ± 17.2^*	14	69.8 ± 15.1	90.0 ± 8.1^*	17
Passive ROM of knee extension	−4.8 ± 3.7	−0.8 ± 2.9^*	14	−4.6 ± 4.1	−1.6 ± 2.6^*	17
Barthel Index	72.5 ± 4.1	92.1 ± 3.5^*	14	68.9 ± 7.2	88.0 ± 7.3^*	17
Tinetti scale	18.1 ± 3.0	24.5 ± 1.7^*	14	16.5 ± 2.2	21.6 ± 3.1^*	17
Lequesne Index	19.5 ± 2.8	10.5 ± 4.4^*	14	20.3 ± 1.8	13.9 ± 2.6^*	17
(Bellelli et al., 2010)
Tinetti scale	12.7 ± 6.9	22.4 ± 4.0^†	30	15.7 ± 4.7	23.2 ± 1.6	30
Functional status/patients with Parkinson’s disease
(Pelosin et al., 2010)
Tinetti part I (score)	7.9 (1.8)	NR	9	7.6 (0.9)	NR	9

SD: standard deviation; IR: interquartile range; ROM: range of motion; N=Number in each group.

Significant within-group difference, p<0.05.

†

Significant between-group difference, p<0.05.

Walking ability

This variable was measured through the timed up and go test, the 10-meter walking test, the 6-minute walk test, and the figure-of-8 walk test.

The timed up and go (TUG) test is used for dynamic balance assessment which records the time taken to rise from a chair (height: 50 cm), walk 3 m, turn around a marker, walk back to the chair, and sit down (Bang et al., 2013).

Eight studies assessed the TUG test as one of their primary outcome measures (Bang et al., 2013; Kim and Lee, 2013, 2015; Park and Hwangbo, 2015; Motaqhey et al., 2015; Pelosin et al., 2010; Santamato et al., 2015; Park et al., 2014b), from which, five included stroke patients, two included patients with Parkinson’s disease, and one study recruited patients who had undergone knee arthroplasty. Four out of five studies, which included chronic stroke patients, found significant improvements in the TUG test in AO group compared with the control group (Bang et al., 2013; Kim and Lee, 2013; Park and Hwangbo, 2015; Motaqhey et al., 2015) (Table 7). Motaqhey et al. (2015) found more favorable improvements in women compared with men who received AOT. However, Pelosin et al. (2010) and Santamato et al. (2015) found no improvement in the TUG test following AOT in Parkinson’s disease (Table 7). Furthermore, no significant difference was found between the two groups of patients with knee arthroplasty in terms of TUG results (Park et al., 2014b) (Table7).

Table 7.

Summary data (mean± SD/median (IR), sample size) and significance of within and between-group differences of average scores of walking ability and balance outcome measures following action observation training.

	Action observation group			Control group
Walking ability and balance/stroke patients	Pre-intervention	Post-intervention		Pre-intervention	Post-intervention
(Bang et al., 2013)
TUG (s)	17.51±1.6	12.94±2.17^*†	15	18.01±1.46	15.66±1.75^*	15
10MWT (m/s)	0.65±0.63	1.01±0.12^*†	15	0.66±0.55	0.82±0.11^*	15
6MWT(m)	153.7±21.2	243.3±20.02^*†	15	159.9±29.6	192.8±25.8^*	15
(Kim and Lee, 2013)
TUG (s)	34.33 ± 13.33	23.71 ± 10.39^*†	9	35.78 ± 12.84	33.33 ± 12.10^*	9
WAQ (score)	39.22 ± 6.74	47.00 ± 6.69^*	9	35.67 ± 7.00	41.78 ± 8.30^*	9
FAC (score)	2.56 ± 0.73	3.22 ± 0.97^*	9	2.11 ± 0.60	2.56 ± 1.01	9
(Park and Hwangbo, 2015)
TUG (s)	20.0±3.5	15.7±3.3^*†	20	20.5±3.5	17.7±3.2^*	20
10-M WT (s)	19.1±1.7	14.6±2.3^*†	20	18.6±2.8	15.5±2.3^*	20
Sway area (mm²)	437.9±381.2	167.5±166.2^*	20	349.7±407.6	349.7±407.6^*	20
Sway speed (mm/s)	0.8±0.3	0.5±0.2^*†	20	0.7±0.2	0.6±0.2^*	20
Limit of stability (mm²)	4,235±4,644.7	7,064±5,014^*†	20	2,361±2,299	3,002±2,629^*	20
(Motaqhey et al., 2015)
TUG (s)	24.8 ± 1	21.5 ± 1.2^*†	6	25.3 ± 1	23 ± 1.4^*	6
6MWT(m)	117.8 ± 6	138.6 ± 4.3^*†	6	104.2 ± 14	120.8 ± 10^*	6
BBS	47.6 ± 2	50.8 ± 2.5^*†	6	46.6 ± 1.6	48.5 ± 1.7^*	6
(Kim and Lee, 2015)
TUG	18.81±9.25	16.29±9.47	6	19.91±22.20	16.91±7.27	6
DGI	18.50±4.80	18.00±6.19	6	13.66±5.68	17.40±3.78	6
(Park et al., 2014a)
10MWT(s)	28.9 (16.1–55.7)	23 (14.5–36.7)^*†	11	22.2 (13.3–48.6)	19.8 (13.9–38.7)	10
F8WT (time)	22.2 (15.7–47.6)	20 (14.1–31.5)^*†	11	23.8 (15.7–47)	19.9 (15.5–40.2)	10
F8WT (steps)	31 (23–35)	26 (20–30)^*†	11	31 (24.5–34.5)	28.5 (23–32.5)	10
DGI	11 (8–17)	17 (12–18)^*†	11	13.5 (7.5–19)	13 (9–15.75)	10
(Park et al., 2017)
10MWT (m/s)	0.43 ± 0.19	0.60 ± 0.33	12	0.64 ± 0.35	0.70 ± 0.39	13
CWT (s)	899.8 ± 360.78	748.4 ± 352.75	12	729.3 ± 439.04	796.38± 587.39	13
ABC (scores)	57.74 ± 22.76	63.99 ± 20.96^*†	12	44.91 ± 31.42	45.62 ± 32.43	13
(Song and Lee, 2016)
10-M WT (sec)	19.35±4.5	13.61±3.26^*†	8	18.80±4.35	15.96±3.33^*	8
6MWT(m)	212.25±82.94	245.25±82.48^*	8	211.5±59.26	231.62±63.66^*	8
DGI	17.13±4.05	20.50±3.16^*†	8	15.87±2.64	16.87±3.22^*	8
(Park and Kang, 2013)
Weight distribution (%), anterior-posterior	36.15±6.56	40.35±4.28^*	10	37.23±10.37	40.16±6.97^*	10
Weight distribution (%), right-left	29.87±5.48	32.57±4.49^*	10	28.26±9.20	28.73±9.05	10
Stability index	23.23±4.54	19.11±5.32^*	10	22.45±6.55	21.69±5.96	10
(Lee et al., 2017)
Anteroposterior Balance Index	1.3 ± 1.0	0.6 ± 0.5^*	12	0.8 ± 0.5	0.9 ± 0.7	11
Mediolateral Balance Index	1.4 ± 1.7	0.8 ± 0.6	12	0.5 ± 0.3	0.5 ± 0.3	11
Fall risk	2.8 ± 1.4	2.3 ± 0.5	12	2.6 ± 0.6	2.1 ± 1.0	11
Walking ability and balance, patients with Parkinson’s disease
(Pelosin et al., 2010)
TUG (s)	32.5 (4.0)	NR	9	28.9 ± 7.3	NR	9
10MWT(s)	15.3 (5.3)	NR	9	15.6 ± 5.0	NR	9
BBS	39.8 ± 3.6	NR	9	40.6 ± 4.6	NR	9
FOG -Q (score)	18.6 ± 1.9	14.1 ± 2.8^*†	9	17.8 ± 2.0	16.4 ± 2.5^*	9
(Santamato et al., 2015)						9
TUG (s)	15.3 ± 4.2	14.1 ± 4.1	15	NA	NA	0
10MWT(s)	16.1 ± 3.3	15.0 ± 3.3	15	NA	NA	0
BBS	42.9 ± 6.9	44.8 ± 7.0	15	NA	NA	0
(Agosta and Sarasso, 2016)
10MWT normal comfortable speed (s)	9.0 ± 1.7	8.2 ± 1.1^*	12	8.0 ± 1.8	7.2 ± 1.2^*	13
10MWT maximum speed (s)	6.56 ± 1.77	6.0 ± 1.4^*	12	6.3 ± 1.4	5.6 ± 1.0^*	13
FOG-Q (score)	11.7 ± 2.9	9.7 ± 3.4^*	12	12.6 ± 3.8	10.9 ± 3.0^*	13
BBS	50.9 ± 3.8	53.6 ± 2.6^*	12	52.2 ± 4.3	54.4 ± 2.4	13
Walking ability and balance, orthopedic patients
(Park et al., 2014b)
TUG (s)	42.23 ± 11.03	18.11± 3.06	9	38.01 ± 9.45	18.31± 4.48	9

SD: standard deviation; IR: interquartile range; ROM: range of motion; N=Number in each group, TUG: Timed Up and Go; 10MWT: 10-Meter Walk Test; 6MWT: 6 Minute Walk Test; WAQ: Walking Ability Questionnaire; FAC: Functional Ambulation Category; BBS: Berg Balance Scale; DGI: Dynamic gait index; F8WT: Figure-of-8 Walk Test; CWT: community walk test; ABC: activity-specific balance confidence scale; FOG-Q: freezing of gait-questionnaire.

* Significant within-group difference, p<0.05.

† Significant between-group difference, p<0.05.

Ten-meter walking test (10-MWT) is a test to record the walking speed. Subjects are asked to walk 12m at their comfortable speed, and the time is recorded for 10m walked using a stopwatch (Park et al., 2017).

Eight studies used the 10-MWT to evaluate the walking speed. From those studies, five were on stroke patients (Song and Lee, 2016; Park et al., 2014a, 2017; Park and Hwangbo, 2015; Bang et al., 2013) and three on patients with Parkinson’s disease (Agosta and Sarasso, 2016; Pelosin et al., 2010; Santamato et al., 2015). All studies on stroke patients showed significant improvement in the 10-MWT in AO group (Bang et al., 2013; Park et al., 2014a, 2017; Park and Hwangbo, 2015; Song and Lee, 2016) (Table 7). Furthermore, none of the studies on patients with Parkinson’s disease found differences between the AO and control group for the studied variable.

6-minute walk test (6-MWT) was used by Bang et al. (2013) and Song and Lee (2016) to measure the effect of AOT on patients’ endurance as an indicator of walking ability.

Action observation before treadmill training resulted in a significant improvement in endurance test compared with landscape watching group (Bang et al., 2013), while watching and doing treadmill training at the same time did not yield a significant difference in 6MWT compared with watching unrelated videos among chronic stroke patients (Song and Lee, 2016) (Table 7).

Figure-of-8 walk test (F8W)

The patients’ ability for walking in different routes, including straight and curved ones, was measured by the F8W test only in one study recruiting chronic stroke patients which found a significant improvement in AO compared with the control group in this walking ability (Park et al., 2014a) (Table 7).

Community walking and ambulation measured by Walking Ability Questionnaire and Functional Ambulation Category, respectively, used in only one study and did not show any statistically significant difference between the AO and physical training groups in patients with stroke (Kim and Lee, 2013) (Table 7). However, community walk test employed by Park et al. (2017) revealed a more favorable outcome after intervention only in the AO group amongst patients with chronic stroke (Table 7).

Freezing of gait (FOG) is a specific feature of gait in patients with Parkinson’s disease. Two out of four studies on patients with Parkinson’s diseases included in this review considered FOG as one of their outcome measures. The between-group difference in the number of FOG episodes was significant in the study by Pelosin et al. (2010). Agosta and Sarasso (2016) found reduced severity of FOG in both groups after treatment, though the difference between groups failed to reach the significance level (Table 7).

Balance

13 studies assessed balance defined through the stability and weight bearing (Park and Kang, 2013; Park et al., 2014a, 2017; Park and Hwangbo, 2015; Song and Lee, 2016; Lee et al., 2017; Motaqhey et al., 2015; Villafane et al., 2016, 2017; Pelosin et al., 2010; Santamato et al., 2015; Agosta and Sarasso, 2016; Kim and Lee, 2015). Four studies used the berg balance scale (BBS) to assess balance following AOT that three studies included patients with Parkinson’s disease (Agosta and Sarasso, 2016; Santamato et al., 2015; Pelosin et al., 2010). Two of three studies on Parkinson’s disease failed to demonstrate any significant difference between the AO and control group in terms of the BBS score (Pelosin et al., 2010; Santamato et al., 2015) (Table 7). Meanwhile, Agosta et al. (2017) found a favorable effect on balance deficits in the AO group which maintained for a long time after training (Table 7). The only study on chronic stroke patients found more improvement in the BBS score of the patients following AO compared to the control condition with more improvements in female than male group (Motaqhey et al., 2015) (Table 7).

Dynamic gait index (DGI) was used in three studies that recruited chronic stroke patients (Park et al., 2014a; Song and Lee, 2016; Kim and Lee, 2015). Two studies found more improvements in the AO group compared with the control group (Park et al., 2014a; Song and Lee, 2016) (Table 7). Meanwhile, a report by Kim and Lee did not highlight any significant difference in DGI scores following AO (Kim and Lee, 2015) (Table 7).

There was no difference between the AO and control groups in any of the balance variables measured by the Biodex Balance scale, postural stability, and fall risk (Lee et al., 2017) (Table 7).

On the other hand, the weight-bearing and stability index was measured in one study, and its results showed significant differences in the left-right weight-bearing and stability index between the two groups (Park and Kang, 2013) (Table 7).

Significant differences in the sway speed and total length of the limit of stability were observed between groups after experiments, while no difference was found in the sway area in a recent report (Park and Hwangbo, 2015) (Table 7).

Activities-specific balance confidence scale in stroke subjects, which was measured by (Park et al., 2017), indicated a pronounced post-intervention improvement in the experimental but not control group.

In recent studies, the functional status and gait assessed in postoperative orthopedic conditions by different assessment tools, including Barthel index, Tinetti scale, and Lequesne index failed to show a significant impact of AOT rather than conventional therapy in groups of patients with knee or hip arthroplasty (Villafane et al., 2016, 2017) (Table 6). Further, no between-group difference was found in the Tinetti score in patients with Parkinson’s disease after AO (Pelosin et al., 2010). However, according to Bellelli et al. (2010), postoperative patients in the AO group achieved higher post-intervention Tinetti scores than controls (Table 6).

The FIM motor sub-score measured in post-operative condition by Bellelli et al. (2010) was found to be significantly higher in the experimental group than controls.

Discussion

The present study aimed to review the available evidence concerning the benefits of AOT on different aspects of gait and balance control in patients with neurological and musculoskeletal disorders. A total of 19 studies were included in this review.

Spatiotemporal and kinematic parameters of gait

Musculoskeletal diseases such as knee osteoarthritis commonly have adverse effects on walking patterns. In some instances, even after the surgery, the impaired spatiotemporal patterns and asymmetrical gait persist (Levinger et al., 2009). Disturbed spatial and temporal aspects of gait are also common among patients with neurological disorders specially stroke and Parkinson’s disease (An et al., 2017; Fling et al., 2018). Given the variety of conditions influenced, improving spatiotemporal gait parameters might be considered as one of the main goals in the rehabilitative programs.

In the present review, spatiotemporal parameters were evaluated in 5 out of 19 studies which their quality scores varied from 55.5 to 74.7 (Table 1) (Jaywant et al., 2016; Park et al., 2014a, 2017; Kim and Kim, 2012; Kim and Lee, 2013). Four studies recruited stroke patients from which three found more improvement of different aspects of spatiotemporal parameters following AO in addition to physical training (Park et al., 2017; Kim and Kim, 2012; Kim and Lee, 2013). Only one study on stroke patients did not find any significant between-group differences (Park et al., 2014a). A possible explanation for this controversy might be the number of treatment sessions and the time spent on AO in the study by Park et al. (2014a) compared to the three other studies. Although the number of treatment sessions has not been mentioned in the study by Kim and Kim (2012), we must be cautious in our reasoning. The only study on Parkinson’s disease did not find any significant improvement in the AO group than the control group (Jaywant et al., 2016). However, compared to studies on stroke patients, it seems that the method of intervention and also the duration of treatment in the study by Jaywant et al. (2016) might have led to this result. In this study, AO was done at home without any other physical training and just for 1 week. According to these results, it seems that stroke survivors are the patients who might benefit the most recovery of spatiotemporal gait parameters by additional AOT. However, more studies on stroke and other neurodegenerative diseases such as Parkinson’s disease are required for supporting this deduction.

The active and passive range of motion measured in two studies on post-operative patients were not significantly affected by AOT (Villafane et al., 2016, 2017). The results of these two studies recruiting the patients with the same disorder support each other and suggest that the patients with orthopedic problems might not drive considerable benefits of AO in improving joint range of motion.

Recovery of Locomotor capacity after AOT

Walking ability in terms of walking speed, endurance, and the ability to deal with different routes, surfaces, and community situations had been considered in about half of the reviewed studies. The ability of non-stop walking for at least half a kilometer with reasonable and safe speed for crossing the road is supposed to be crucial for community ambulation after stroke. However, a small percentage of people meet this criterion even after finishing their rehabilitation program (Dean et al., 2001). Walking speed is the most evaluated aspect of walking ability measured by 10-MWT in seven studies with the quality scores varying from 55.5% to 81.48% (Pelosin et al., 2010; Bang et al., 2013; Santamato et al., 2015; Agosta and Sarasso, 2016; Park et al., 2014a, 2017; Song and Lee, 2016; Park and Hwangbo, 2015) (Table 1). The results show that patients with stroke experienced a better recovery of gait speed following AO accompanying physical training compared with physical therapy alone (Bang et al., 2013; Park et al., 2014a, 2017; Song and Lee, 2016; Park and Hwangbo, 2015). In addition to speed, people who watched exercise before executing represented better endurance during walking (Bang et al., 2013) while watching and executing simultaneously did not have a significant effect on the patient’s endurance (Song and Lee, 2016). This discrepancy might be attributed to the changes in motor representation and establishment of a motor memory induced by AO before execution, which has been shown in a previous study (Stefan et al., 2005). So, the formation of a memory of an action preceding its execution may prepare the neural circuits for better acquisition of a skill which might be less prominent in doing and acting at the same time. On the other hand, considering the intervention setting of these two studies, differences in the time spent on exercise and the number of sessions as well as the number of subjects recruited may be the possible causes of this variation. However, the results of 10MWT supported each other in both studies (Bang et al., 2013; Song and Lee, 2016). Therefore, it seems that further studies with the same intervention are required to get a better understanding of this scope. Furthermore, AO resulted in a better ability to deal with different community environment measured by F8W (Park et al., 2014a) and community walk test (Park et al., 2017). However, the results of the study by Kim and Lee (2013) do not support the two other studies regarding walking ability and functional ambulation. This disparity might be due to the different measurement tools.

Despite promising outcomes on locomotor capacities of stroke patients, Parkinsonians did not achieve much benefit from AO rather than that of physical therapy for walking speed (Pelosin et al., 2010; Santamato et al., 2015; Agosta and Sarasso, 2016). However, AO as a supplementary treatment affected the freezing of gait, a specific feature of Parkinson’s disease, significantly in the study by Pelosin et al. (2010). Also, the other study by Agosta and Sarasso (2016) found the same trend in reduction of FOG severity but it did not reach the significance level. Given applying the same intervention method and even a larger sample size in the study by Agosta and Sarasso (2016) justifying this disparity is somehow difficult. The mean age of subjects assigned to each group might be suggested as the probable reason for this difference. In the study by Pelosin et al. (2010) subjects in the experimental group were younger than that of control, while in the other study (Agosta and Sarasso, 2016) younger subjects were allocated to the control group.

Abnormally high beta power recorded from subthalamic nucleus (STN) during “off” condition of patients with Parkinson’s disease has been suggested as a cause of bradykinesia (Alegre et al., 2010). On the other hand, in a more recent study, the relationship between FOG and increased high beta band power in positive-FOG patients has been reported. The FOG recovery following dopaminergic treatment has been associated with a significant decrease in this brain oscillatory activity (Toledo et al., 2014). Movement observation has also resulted in a beta reduction-similar in pattern but smaller and shorter than that of movement execution in STN of patients with Parkinson’s disease. It is known that basal ganglia, a set of structures involved in facilitation/inhibition of motor programs, are connected to the cortical motor areas through STN. Therefore, changes in cortical activity induced by mirror neurons during AO can be delivered to basal ganglia and specially STN through this pathway and result in beta band reduction and FOG recovery (Alegre et al., 2010).

Balance and postural control

Balance and stability measured by different assessment tools have been reported in 13 articles included in this review (Agosta and Sarasso, 2016; Pelosin et al., 2010; Villafane et al., 2016, 2017; Kim and Lee, 2015; Motaqhey et al., 2015; Lee et al., 2017; Song and Lee, 2016; Park et al., 2014a, 2017; Park and Hwangbo, 2015; Park and Kang, 2013; Santamato et al., 2015) and their quality scores were between 51.85% to 81.48% (Table 1). Although more improvement was observed following AO especially for women with stroke, the results of the berg balance scale (BBS) did not show the significant priority of AO than physical therapy. Balance improvement score measured by BBS met the significance level in just one study recruiting patient with Parkinson’s disease (Agosta and Sarasso, 2016) while the other two studies observed no between-group difference. This variance might be related to the different sample size evaluated in three studies. Agosta and colleagues (2017), who recruited more patients, succeeded to show the efficacy of AO, while Pelosin et al. (2010) did not observe the same result even by using the same intervention method and rehabilitation program. The possible explanation might be the lower number of patients they recruited in their study. The other study on only one group of patients without any control group also failed to show the post- to-pre-treatment amelioration of stability in patients with Parkinson’s disease. Although they implemented a 2-month intervention, their small sample size might justify the results (Santamato et al., 2015). Two out of four studies that used DGI in stroke patients reported more improvement in the experimental group (Song and Lee, 2016; Park et al., 2014a). Considering the study protocols, the fewer subjects and less time spent on observation might be suggested as the cause of disparity between the study by Kim and Lee (2015) and two other studies. Other aspects of balance, including side to side weight bearing, stability index, sway speed, as well as the limit of stability, have been reported to improve more following AO just in two studies (Park and Kang, 2013; Park and Hwangbo, 2015). It has also been reported that stroke patients had experienced more balance confidence in specific activities following AO (Park et al., 2017).

The adult brain has a highly dynamic nature that allows the experienced-based reorganization of the cortical connections, which is important for learning and recovery following neural injuries (Mulder, 2007). Relearning strategies, as well as the adaptation of lesion, are some mechanisms suggested for recovery of motor deficits after stroke. Traditional approaches for motor recovery after stroke are based on this idea that repetitive motor movement by affected limb leads to neuronal plasticity resulting in faster recovery (Ertelt et al., 2007). Evidence for the presence of a neuronal system with the ability of activation during both observation and execution of the same actions comes from several neuroimaging studies. The studies show that this so-called mirror neuron network, consisting of cortical sensory-motor areas, produces neuronal excitability and muscle activation pattern during AO, which is very similar in terms of quality and specificity to that of action execution but smaller in size. Changes in motor representation during AO have been suggested to result in the formation of a motor memory, which seems might facilitate the acquisition of motor skills (Fadiga et al., 1995; Stefan et al., 2005).

The result of the study performed by Ertelt et al. (2007) on the recovery of upper limb motor function in a group of stroke patients provides supportive evidence for the effectiveness of AOT after stroke. Larger post-treatment activation was found in the AO group in brain areas building up the mirror neuron system, including sensorimotor networks, compared with the control group. Furthermore, the observed increase in neuronal activity was in line with functional motor improvement (Ertelt et al., 2007).

Based on the aim of this study, the effect of AOT on lower limb function of patients with neurological or musculoskeletal disorders was assessed. The studies on stroke patients revealed significant improvement in spatiotemporal parameters, walking speed, and balance following AOT. Patients with Parkinson’s disease did not reach better outcomes with AOT in spatiotemporal parameters or walking ability tests compared with the control group. However, FOG, especially measured for this group of patients, seems to be influenced favorably by AOT.

The few studies on musculoskeletal disorders discussed here did not assess most of the primary outcome measures considered in this review. However, the results suggest a relatively better improvement in the patients’ mobility and locomotion in the AO group.

Rehabilitation is of great importance in improving the independence and quality of life of people with acquired neurological disorders. The beneficial effects of current multidisciplinary rehabilitation approaches are well established for people with acquired neurological conditions such as stroke (Clark et al., 2019).

Occupational therapy is considered a principal part of the multidisciplinary rehabilitation of stroke patients, which improves outcomes in everyday life occupations such as activities of daily living (ADL) and participation (Landi et al., 2006). Nevertheless, even after completing rehabilitation courses, the residual gait disturbance in these patients limits their ability to perform daily activities (Rea et al., 2014). Therefore, the interest in exploring novel rehabilitation technologies to enhance the positive effects of conventional therapies for reducing neurological disability and improving function is increasing (Clark et al., 2019). Accordingly, research-driven evidence regarding the effectiveness of AOT for stroke patients suggests that this neuromodulatory technique may enhance the efficacy of conventional motor rehabilitation, including occupational therapy, on the gait recovery of stroke patients.

In conclusion

The results of the papers included in the present review suggest that the AO training accompanying physical training such as occupational therapy has the potential to enhance the beneficial effects of physical training on spatiotemporal gait parameters and the walking ability of patients with stroke. Furthermore, patients with Parkinson’s disease can at least achieve a better functional result following AO accompanying conventional physical training in their FOG severity.

Key findings

Patients with a stroke can probably take more advantage of the beneficial effects of action observation therapy accompanied by conventional physical training such as occupational therapy for gait improvement than those with Parkinson’s disease or patients with neuromuscular disorders.

What the study has added

This study highlights the need for strategies to address the knowledge-practice gap, including evidence-based guidelines. Closer collaboration between occupational therapists and school staff could increase understanding of roles and highlight the unique occupational therapy contribution.

Footnotes

Author contributions

FSH researched literature and conceived the study. MTK was involved in protocol development. FSH wrote the first draft of the manuscript. All authors reviewed and edited the manuscript and approved the final version of the manuscript.

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

References

Agosta

F, R. G

Sarasso

, et al. (2016) Brain plasticity in Parkinson’s disease with freezing of gait induced by action observation training. Journal of Neurology 264: 88–101.

Alegre

Rodriguez-Oroz

Valencia

, et al. (2010) Changes in subthalamic activity during movement observation in Parkinson's disease: is the mirror system mirrored in the basal ganglia? Clinical Neurophysiology: Official Journal of the International Federation of Clinical Neurophysiology 121: 414–425.

C-M

Son

Y-L

Park

Y-H

, et al. (2017) Relationship between dynamic balance and spatiotemporal gait symmetry in hemiplegic patients with chronic stroke. Hong Kong Physiotherapy Journal 37: 19–24.

Bang

Shin

Kim

, et al. (2013) The effects of action observational training on walking ability in chronic stroke patients: a double-blind randomized controlled trial. Clinical Rehabilitation 27: 1118–1125.

Bellelli

Buccino

Bernardini

, et al. (2010) Action observation treatment improves recovery of postsurgical orthopedic patients: evidence for a top-down effect? Arch Phys Med Rehabil 91: 1489–1494.

Buccino

Arisi

Gough

, et al. (2012) Improving upper limb motor functions through action observation treatment: a pilot study in children with cerebral palsy. Developmental Medicine and Child Neurology 54: 822–828.

Buccino

Gatti

Giusti

, et al. (2011) Action observation treatment improves autonomy in daily activities in parkinson’s disease patients: results from a pilot study. Movement Disorders: Official Journal of the Movement Disorder Society 26: 1963–1964.

Buccino

Vogt

Ritzl

, et al. (2004) Neural Circuits Underlying Imitation Learning of Hand Actions: An Event-Related fMRI Study. Neuron 42: 323–334.

Chatterton

Ewan

Kinmond

, et al. (2008) Observation of meaningful activities: a case study of a personalized intervention on poststroke functional state. Journal of Neurologic Physical Therapy: JNPT 32: 97–102.

10.

Chen

Patten

Kothari

, et al. (2005) Gait differences between individuals with post-stroke hemiparesis and non-disabled controls at matched speeds. Gait Posture 22: 51–56.

11.

Clark

Sivan

O’Connor

. (2019) Evaluating the use of robotic and virtual reality rehabilitation technologies to improve function in stroke survivors: A narrative review. Journal of Rehabilitation and Assistive Technologies Engineering 6: 2055668319863557.

12.

Dean

Richards

Malouin

. (2001) Walking speed over 10 metres overestimates locomotor capacity after stroke. Clinical Rehabilitation 15: 415–421.

13.

Down

Black

. (1998) The feasibility of creating a checklist for the assessment of the methodological quality both of randomised and non-randomised studies of health care interventions. Journal of Epidemiology and Community Health 52: 377–384.

14.

Ertelt

Small

Solodkin

, et al. (2007) Action observation has a positive impact on rehabilitation of motor deficits after stroke. Neuroimage 36: 164–173.

15.

Fadiga

Fogassi

Pavesi

, et al. (1995) Motor facilitation during action observation: a magnetic stimulation study. Journal of Neurophysiology 73: 2608–2611.

16.

Ferrari

Gallese

Rizzolatti

, et al. (2003) Mirror neurons responding to the observation of ingestive and communicative mouth actions in the monkey ventral premotor cortex. European Journal of Neuroscience 17: 1703–1714.

17.

Fling

Curtze

Horak

. (2018) gait asymmetry in People With Parkinson’s Disease is linked to reduced integrity of callosal sensorimotor regions. Front Neurol 9: 215.

18.

Franceschini

Ceravolo

Agosti

, et al. (2012) Clinical relevance of action observation in upper-limb stroke rehabilitation: a possible role in recovery of functional dexterity. A randomized clinical trial. Neurorehabilitation and Neural Repair 26: 456–462.

19.

Ghanjal

Torkaman

Ghabaee

, et al. (2015) The Effect of Action Observation on Weight Distribution and Dynamic Balance Index Improvement in Hemiparetic Patients Based on Mirror Neuron Theory. Journal of Zanjan University of Medical Sciences 23: 77–88.

20.

Ghanjal

Torkaman

Ghabaee

, et al. (2014) Effect of action observation and imitation on improving the functional activities indices in hemiplegic patients based on mirror neurons theory. Journal of Mazandaran University of Medical Sciences 24: 136–150.

21.

Hollman

Kovash

Kubik

, et al. (2007) Age-related differences in spatiotemporal markers of gait stability during dual task walking. Gait Posture 26: 113–119.

22.

Jaywant

Ellis

Roy

, et al. (2016) Randomized controlled trial of a home-based action observation intervention to improve walking in parkinson disease. Arch Phys Med Rehabil 97: 665–673.

23.

Kim

Lee

. (2015) The effect of action observation training on sit to walk with chronic stroke patients. J Kor Phys Ther 27: 413–418.

24.

Kim

Lee

. (2013) Action observation training for functional activities after stroke: a pilot randomized controlled trial. Neurorehabilitation 33: 565–574.

25.

Kim

Lee

, et al. (2014) Differences in Brain Waves of Normal Persons and Stroke Patients during Action Observation and Motor Imagery. J. Phys. Ther. Sci 26: 215–218.

26.

Kim

. (2012) Clinical Feasibility of action observation based on mirror neuron system on walking performance in post stroke patients. J. Phys. Ther. Sci 24: 597–599.

27.

Krebs

Jette

Assmann

. (1998) Moderate exercise improves gait stability in disabled elders. Arch Phys Med Rehabil 79: 1489–1495.

28.

Landi

Cesari

Onder

, et al. (2006) Effects of an occupational therapy program on functional outcomes in older stroke patients. Gerontology 52: 85–91.

29.

LaRocca

(2011) Impact of walking impairment in multiple sclerosis. The Patient: Patient-Centered Outcomes Research 4: 189–201.

30.

Lee

Kim

Lee

. (2017) The effects of action observation training and mirror therapy on gait and balance in stroke patients. The Journal of Physical Therapy Science 29: 523–526.

31.

Levinger

Lai

Begg

, et al. (2009) The application of support vector machines for detecting recovery from knee replacement surgery using spatio-temporal gait parameters. Gait Posture 29: 91–96.

32.

Mattar

Gribble

. (2005) Motor learning by observing. Neuron 46: 153–160.

33.

Merino

Glaser

Gre` zes

, et al. (2005) Action observation and acquired motor skills: an FMRI study with expert dancers. Cerebral Cortex 15: 1243–1249.

34.

Motaqhey

Ghanjal

Mastri Farahani

, et al. (2015) Sex Differences in Neuroanatomy of the Human Mirror Neuron System: Impact on Functional Recovery of Ischemic Hemiparetic Patients. Iran Red Crescent Med J 17: e28363.

35.

Mulder

(2007) Motor imagery and action observation: cognitive tools for rehabilitation. J Neural Transm (Vienna) 114: 1265–1278.

36.

Park

Kang

. (2013) The effects of additional action observational training for functional electrical stimulation treatment on weight bearing, stability and gait velocity of hemiplegic patients. J. Phys. Ther. Sci 25: 1173–1175.

37.

Park

Hwangbo

. (2015) The effects of action observation gait training on the static balance and walking ability of stroke patiens. J. Phys. Ther. Sci 27: 341–344.

38.

Park

Choi

, et al. (2017) Action observation training of community ambulation for improving walking ability of patients with post-stroke hemiparesis: A randomized controlled pilot trial. Clinical Rehabilitation 31: 1078–1086.

39.

Park

Kim

Lee

, et al. (2014a) Clinical feasibility of action observation training for walking function of patients with post- stroke hemiparesis: a randomized controlled trial. Clinical Rehabilitation 28: 794–803.

40.

Park

Song

Kim

. (2014b) The effect of action observation training on knee joint function and gait ability in total knee replacement patients. J Exerc Rehabil 10: 168–171.

41.

Patel

(2017) Action observation in the modification of postural sway and gait: Theory and use in rehabilitation. Gait Posture 58: 115–120.

42.

Pelosin

Avanzino

Bove

, et al. (2010) Action observation improves freezing of gait in patients with Parkinson's disease. Neurorehabilitation and Neural Repair 24: 746–752.

43.

Pirker

Katzenschlager

. (2016) Gait disorders in adults and the elderly. Wiener Klinische Wochenschrift 129: 81–95.

44.

Pizzi

Carlucci

Falsini

, et al. (2007) Gait in hemiplegia: evaluation of clinical features with the Wisconsin Gait Scale. Journal of Rehabilitation Medicine: Official Journal of the UEMS European Board of Physical and Rehabilitation Medicine 39: 170–174.

45.

Porro

Facchin

Fusi

, et al. (2007) Enhancement of force after action observation: behavioural and neurophysiological studies. Neuropsychologia 45: 3114–3121.

46.

Rea

Rana

Lugato

, et al. (2014) Lower Limb Movement Preparation in Chronic Stroke: A Pilot Study Toward an fNIRS-BCI for Gait Rehabilitation. Neurorehabilitation and Neural Repair 28: 564–575.

47.

Rezzan

Marinelli

Marcato

, et al. (2015) Efficacy of Action Observation Pre-operative Training in Functional Recovery after Hip and Knee Prosthesis.

48.

Sale

Masiero

Walter

, et al. (2018) Action observation and gait recovery in subject with Parkinson's disease. A new motor-cognitive combined approach. A preliminary report. Annals of Physical and Rehabilitation Medicine 61: e252.

49.

Santamato

Ranieri

Cinone

, et al. (2015) Postural and Balance Disorders in Patients with Parkinson's Disease: A Prospective Open-Label Feasibility Study with Two Months of Action Observation Treatment. Parkinsons Dis 2015: 902738.

50.

Small

Buccino

Solodkin

. (2012) The mirror neuron system and treatment of stroke. Developmental Psychobiology 54: 293–310.

51.

Small

Buccino

Solodkin

. (2013) Brain repair after stroke--a novel neurological model. Nature Reviews. Neurology 9: 698–707.

52.

Snijders

van de Warrenburg

Giladi

, et al. (2007) Neurological gait disorders in elderly people: clinical approach and classification. Lancet Neurology 6: 63–74.

53.

Song

Y-h

Lee

H-m

. (2016) The effect of treadmill training applied simultaneously with action observation on walking ability in chronic stroke patients. The Journal of Korean Physical Therapy 28: 176–182.

54.

Stefan

Cohen

Duque

, et al. (2005) Formation of a motor memory by action observation. The Journal of Neuroscience: the Official Journal of the Society for Neuroscience 25: 9339–9346.

55.

Takeshi

Atsushi

Hiroaki

, et al. (2015) Effect of action observation therapy on lower-extremity function of chronic stroke patients. Rigakuryoho Kagaku 30: 251–256.

56.

Toledo

Lopez-Azcarate

Garcia-Garcia

, et al. (2014) High beta activity in the subthalamic nucleus and freezing of gait in Parkinson's disease. Neurobiology of Disease 64: 60–65.

57.

Villafane

Isgro

Borsatti

, et al. (2017) Effects of action observation treatment in recovery after total knee replacement: a prospective clinical trial. Clinical Rehabilitation 31: 361–368.

58.

Villafane

Pirali

Isgro

, et al. (2016) Effects of Action Observation Therapy in Patients Recovering From Total Hip Arthroplasty Arthroplasty: A Prospective Clinical Trial. J Chiropr Med 15: 229–234.

59.

Woolf

Pfleger

. (2003) Burden of major musculoskeletal conditions. Bulletin of the World Health Organization 81: 646–656.