Application of action observation therapy in stroke rehabilitation: A systematic review

Abstract Background Numerous studies have described the positive effects of action observation therapy (AOT) on motor recovery among patients with stroke. However, there is no standardized procedure for when and how to intervene with AOT. Objectives Thus, we reviewed and analyzed previous studies to provide a guideline for the application of AOT in stroke rehabilitation. Method We searched PubMed, Cochrane Library, and EMBASE from inception to October 31 2022, using title and abstract search terms of “action observation” and “stroke” or “hemiplegia.” Of 4108 potential articles, 29 articles (sample size = 429 in AOT groups; sample size = 423 in control groups) that met inclusion criteria were included in final analyses. Results The results suggested starting adjuvant AOT > 23 days after stroke onset and conducting 30–40 min/session, 3–5 times/week for at least 4 weeks. Conclusion Based on our results, many factors will impact the effect of AOT on stroke rehabilitation, when to apply (timing) and how to apply (frequency, single, and total duration) should be fully considered when applying AOT as adjuvant therapy in stroke rehabilitation.

Numerous studies have described the positive effects of AOT for motor recovery among patients with stroke (Celnik et al., 2008;Franceschini et al., 2012;Sugg et al., 2015). Previous reviews indicated positive effect of AOT on the rehabilitation of stroke, some reviews using systematic review or meta-analysis, suggested that AOT had a significant effectiveness on improving the upper limb motor functions, walking ability gait velocity, and daily activity performance in patients with stroke (Borges et al., 2018;Peng et al., 2019;Silverio et al., 2022;Zhang et al., 2019). Another review found that the benefit of incorporating AOT training into rehabilitation programs is strongly supported in populations with Parkinson disease and stroke (Ryan et al., 2021).
However, there is no standardized procedure for when and how best to intervene during the treatment of patients with stroke. Thus, the aim of present study was to provide a guideline based on previous studies for the application of AOT in stroke rehabilitation.

Search strategy
We conducted a literature search of PubMed, Cochrane Library, and EMBASE from inception to October 31, 2022, using the following search terms of titles and abstracts: "action observation" and "stroke" or "hemiplegia." Articles in review articles were manually searched for potential inclusion.

Study selection
The studies selected included those with the following criteria: a longitudinal study published in English in a peer-reviewed journal; participants with clinically diagnosed stroke; control group received sham AOT or no intervention; and behavior outcome measures were reported. Studies reporting only results of imaging or brain activity indexes were excluded.

Data extraction
The following data were extracted from the included studies for further analysis: (1) length of time from stroke onset; (2) sample size; (3) AOT protocol, including intervention duration for one session (e.g., 30 min/session), frequency (times/week), and intervention cycle (e.g., 4 weeks); (4) primary outcome measures; and (5) results of the intervention.

RESULTS
The literature search yielded 4108 potential articles. Following a review of titles and abstracts, 1459 duplicated articles and 711 registered trials without specific results were removed. Two authors then examined the remaining articles for adherence to the criteria defined in Section 2.2. A total of 29 studies were included in the final analyses. The scales and tests used to assess each outcome measure in the included studies are given in Table 2.

When to apply adjuvant AOT after stroke onset
Clinical guidelines have suggested implementing rehabilitation as soon as possible after stroke onset (National Stroke Foundation); however, this guideline may not work for adjuvant AOT with conventional therapy. Of 29 included studies, 22 (75.9%) included patients with stroke onset >1 month, and 7 studies (24.1%) included patients with stroke onset within 1 month. For patients starting intervention at the same time after stroke onset, those starting >1 month had better rehabilitation effects than patients who started intervention within 1 month.
For studies with patients starting intervention at > 1 month of stroke onset, approximately 4 weeks' AOT intervention provided more statistically significant changes in most outcome measures compared with the control group. By contrast, studies with patients starting intervention within 1 month of stroke onset found no or only a few significant differences in outcome measures compared with control group. Among the seven studies (Cowles et al., 2013;Franceschini et al., 2012;Hioka et al., 2020;Mancuso et al., 2021;Noh et al., 2019;Sale et al., 2014;Zhu et al., 2015) whose participants started intervention within 1 month of stroke onset, AOT had no or limited effects on stroke rehabilitation  (Demeurisse et al., 1980) and the Action Research Arm Test (Yozbatiran et al., 2008), and the results indicated no statistical significance between the AOT and control groups before vs. after treatment. Noh et al. (2019) also found no significant differences between the AOT and control groups after 2 weeks' intervention (20 min/session, 5 days/week, 2 weeks; 200 min in total) among patients whose stroke onset was an average of 31.36 (± 16.87) days in the AOT group and 22.63 (± 7.00) days in the control group. Some studies found that only a few measured indexes showed significant differences between the AOT and control groups. For example, one study (Franceschini et al., 2012) examined global motor function as measured by the Fugl-Meyer Assessment (Fugl-Meyer et al., 1975) and the Modified Ashworth Scale (Bohannon & Smith, 1987), activities of daily living as measured by the Functional Independence Measure (Mathiowetz et al., 1985),  Zhu et al. (2015) found that the scores for all the measured indices, which included the Fugl-Meyer Assessment, Barthel Index, and Modified Ashworth Scale, were significantly better after 8 weeks' intervention (30 min/session, 6 times/week; 1440 min in total) in the AOT group compared with controls (for stroke onset times; see Table 1). However, another study showed that for patients whose stroke onset was a mean of 16.6 ± 3.9 days, even when treatment was extended to 12 weeks (30 min/session, 5 days/week for 12 weeks; 1900 min in total), a significant improvement was observed in the AOT group compared with the control group only for the 10-m walking test. In addition, this difference was found only for patients who could walk independently. No significant differences were found between the AOT and control groups for any of the other outcome measures assessed, including the National Institutes of Health Stroke Scale, Brunnstrom recovery stage, modified Rankin Scale, and Timed Up and Go test (Hioka et al., 2020). By contrast, when stroke onset was >23 days, the results were more favorable for AOT. Hence, when using adjuvant AOT with conventional therapy in stroke rehabilitation, we suggest starting AOT 23 or more days after stroke onset.
TA B L E 2 Scales and tests used to assess the indicated outcome measures in the included studies.
As described in Section 3.2, for patients whose stroke onset was >1 month prior to starting AOT, patients in the AOT group had greater improvement in most outcome measures than patients in the control group. However, further analyses showed that in one study implementing AOT for 60 min per session (5 times/week, 3 weeks; 900 min in total), only Functional Independence Measure scores had a greater improvement in the AOT group than in the control group, with no significant differences between these groups for the Fugl-Meyer Assessment, Box and Block test, and Stroke Impact Scale scores (Hsieh et al., 2020). Patients with stroke have difficulty maintaining concentration beyond 20 min (Simmons et al., 2008), which may explain why AOT for 60 min per session across 3 weeks did not result in beneficial effects. Therefore, when employing AOT in stroke patients, we should not only consider a minimum single session duration to ensure the efficacy, but also should be aware the single session duration not be too long which may beyond patients' tolerance. Our results do not clearly indicate an optimal AOT session duration. However, combined with the characteristics of typical patients with stroke and the negative results of the study by Hsieh et al., our results suggest 30-40 min per session may be most beneficial.

3.4%).
Given that most studies conducted AOT either 5 times/week or 3 times/week, we screened for studies with the same amount of time per session and the same number of weeks of intervention. We found that although 5 times/week (30 min/session, 4 weeks) was most commonly used, some studies showed that AOT conducted 3 times/week
After excluding seven studies in which patient stroke onset was within 1 month of the start of AOT intervention, 4 weeks (13 studies, 59.1%) remained the most common, followed by 6 weeks (3 studies,  . However, another study using 3 weeks' AOT (60 min/time, 5 times/week; 900 min in total) found significant differences between AOT and control groups only for Functional Independence Measure scores, but not for Fugl-Meyer Assessment, Box and Block test, and Stroke Impact Scale scores (Hsieh et al., 2020).
Therefore, although additional corroborating evidence is required to conclusively determine the optimal number of weeks that patients with stroke should receive AOT, we suggest that at least 4 weeks of treatment should be assessed in future studies.

DISCUSSION
Most patients with stroke experience motor deficit (Kwakkel et al., 2003), some of them with severe motor deficit had difficult in receiving conventional therapy such as constraint-induced movement therapy (Yavuzer et al., 2008). Thus, alternative therapies are needed to overcome this limitation. AOT is one of the novel approaches whose efficacy were approved by numerous studies (Buccino, 2014;Deconinck et al., 2015).
AOT was developed based on the neural substrate-the mirror neuron system (MNS), which was originally discovered in the monkey premotor and parietal cortex (Rizzolatti & Fogassi, 2014;. Recent works suggested a comparable, distributed mirror neuron system in the human brain that is active during both action observation and execution. Neural activity associated with action observation includes visual processing areas (occipital, temporal, and parietal cortices) and the core mirror neuron system, which comprises motor-related brain regions, including the pars opercularis of the inferior frontal gyrus (IFG), ventral premotor cortex (PMv), and rostral IPL (Grezes &Decety, 2001;Rizzolatti & Craighero, 2004).
A large body of behavioral, neurophysiological, and brain imaging researched indicated that when we observe actions, our motor system simulates those action and increased corticospinal facilitation specific to muscles used to execute the observed actions was found (Urgesi et al., 2006). Moreover, Gazzola and Keysers (2009) (Oh et al., 2019).
Third, first-person versus third-person perspective should be considered: First-person perspective action observation training was not only more effective in improving upper limb function and activities of daily living than third-person perspective action observation training (Yu & Park, 2022) but also had higher localized and selective cerebral activation (Watanabe et al., 2011). The distinction between first-person and third-person perspective has been described elsewhere (Magill, 1998).
Besides, with the emerging of virtual reality-based (VR) technology, VR rehabilitation treatments have been introduced to AOT. Several studies have approved the effectiveness of VR-based rehabilitation in improving the upper limb motor function and daily life activity in stroke patients (Laver et al., 2017;Saposnik et al., 2010;Turolla et al., 2013).
Hence, a combined rehabilitation treatment of AOT and VR-based therapy would be recommended in the future.
There are several limitations that should be considered when interpreting the results of the present study. First, we only analyzed the behavior measures, the future studies should further analyze the brain activity measures to better verify the effectiveness of AOT. Second, we did not consider the impact of duration of the observed video, future study should take this factor into consideration. However, most of the included studies in present review considered the particular situation of stroke patients; they used short-time videos (e.g. 3 min/video) and would ask the patients to take a rest between videos.

CONCLUSIONS
The aim of present systematic review was to offer a guideline on optimal times after stroke onset to begin adjuvant AOT and to provide intervention protocol suggestions as well as some tips on the contents of the AOT videos. Most studies used adjuvant AOT with conventional therapy for rehabilitation of patients with stroke. Our results suggested that adjuvant AOT should begin at least 23 days after stroke onset. The recommended adjuvant AOT protocol is 30-40 min/session, 3-5 times/week for at least 4 weeks. AOT video clips with rhythmic auditory stimulation provided better results than those with only visual stimulation. The use of videos with functional AOT rather than general AOT content and presented from the first-person perspective rather than the third-person perspective was associated with better results.

CONFLICT OF INTEREST STATEMENT
The authors declare no conflicts of interest.

DATA AVAILABILITY STATEMENT
The data that support the findings of this study are available from the corresponding author upon reasonable request.