View all hidden authors and organizations

Contributed by Giacomo Rizzolatti, September 29, 2021 (submitted for review on December 23, 2020; reviewed by Ferdinand Binkofski, Günther Knoblich, and Steven Small)

In several clinical situations, especially those related to orthopedic trauma or specific injuries of the peripheral nervous system, patients may experience a period of non-use of the limbs, which has a detrimental cascading effect on the cortical motor tissue and ultimately on sports performance. . During periods when the limbs are not in use, treatment based on motion observation may be suitable for stimulating the motor system through a mirror image mechanism. Using short-term fixation in healthy volunteers, our research shows that observation of movements during fixation limits the movement changes caused by not using the limbs. In view of the protective effect of movement observation on the decline of exercise ability, it represents an effective tool for early intervention during the period of limbs not being used, thereby reducing the burden of further exercise rehabilitation.

There is a wealth of clinical evidence that observing the normally performed actions can promote the recovery of the corresponding actions performed by patients with motor deficits. In this study, we evaluated the ability of action observation to prevent the decline of healthy individuals' exercise ability after upper limb immobilization. To this end, the upper-limb kinematics of healthy participants were recorded, and they performed 3 grasping movements before immobilization, and performed the same movements 16 hours after immobilization. Participants were subdivided into two groups; the experimental group observed that during the fixation, they performed the same stretch and grasp before fixation, while the control group observed the natural scene. After the bandage was removed, the experimental group had less dyskinesia when performing hand-out movements. These findings support the hypothesis that observing movements through the mirror mechanism can prevent the decline in motor performance caused by the non-use of the limbs. From this perspective, action observation therapy is a promising tool that can predict the start of rehabilitation in clinical situations involving limbs not being used, thereby reducing the burden of further rehabilitation.

There is a wealth of clinical evidence that observing the normally performed actions can promote the recovery of the corresponding actions performed by patients with motor deficits. This procedure is based on activating the motor system through the mirroring mechanism (1, 2), called Action Observation Therapy (AOT) (3, 4). The effectiveness of AOT in exercise recovery has been proven in a variety of clinical diseases, including stroke (5⇓ –7), Parkinson’s disease (8⇓ –10), multiple sclerosis (11) and cerebral palsy (12⇓ ⇓) ⇓ –16), as well as orthopedic trauma patients and postoperative patients (17⇓ –19).

In recent decades, researchers have developed concealed movement methods based on movement observations other than AOT. Such methods include mirror therapy (20, 21), which can improve symptoms caused by missing or changed feedback on the affected side of the body [for example, phantom pain in arm amputees (22, 23)], and can also enhance stroke Patients with later motor function (24⇓ –26). Recently, non-invasive brain stimulation techniques, such as transcranial magnetic stimulation, transcranial direct current stimulation, and peripheral electrical stimulation, have been used to promote the exercise recovery of the nervous system (27⇓-29) and orthopedic patients (30). When tested in conjunction with action observation-based interventions, these methods have demonstrated the ability to enhance treatment effects (31, 32).

In some of the above clinical situations, especially those involving orthopedic trauma or affecting the peripheral nervous system, the patient may experience a period of limb inactivity. It has been proved that the non-use (or disuse) of the limb will lead to the reduction of the size and excitability of the cortical representation of the fixed limb, which will gradually lead to the appearance of maladaptive plastic changes and movement changes (33⇓ –35), which will interfere with the rehabilitation result. In this case, when physical therapy is not applicable, action observation is an effective treatment option. The purpose of this study is to determine whether AOT administered during a fixed period can limit the gradual decline of exercise capacity-researchers have not yet determined this effect.

To this end, short-term immobilization (STI) was performed on healthy volunteers. This program is often used to simulate neurophysiological changes that cause movement disorders in injured persons (reviews in refs. 36 and 37) because it minimizes the influence of confounding variables (e.g., braking duration, braking cause, related Pain, potential comorbidities), thus isolating the effects of decreased activity on neurophysiological processes. In addition, the use of healthy volunteers can compare the performance of the subjects before and after fixation, and due to the sudden nature of the injury, it is almost impossible to perform the same procedure with the clinical population.

Participants were subdivided into two groups: the group receiving AOT and the group receiving control stimuli. In both groups, the upper-limb kinematics of goal-directed movement was tested before and after fixation. We assessed whether subjects who received AOT had better retention of athletic performance after fixation compared to the control group (CTRL) group, and determined which aspects of the athletic tissue were most affected by AOT.

The results of this study may lead to the use of AOT during a fixed period in a series of clinical situations where the patient’s movement is temporarily blocked, which is conducive to the start of early treatment and may limit the degree of motor deficits for later recovery.

In order to evaluate the protective effect of AOT on the dyskinesia that usually occurs after fixation, we recorded the upper limb kinematics of a group of healthy volunteers before and after arm fixation (16 hours). Participants were asked to perform three hand-out and grasping movements, which were differentiated according to the position of the target object, as follows: 1) located in front of the subject's shoulder height (A-Low), 2) located in front of the subject The height of the head (A-High), and 3) the horizontal (L-Low) of the subject's shoulder height (Figure 1A). During the fixation process, half of the participants (AOT group) repeatedly observed and imagined the same actions performed before, while the other half (CTRL group) observed the natural scene at the same time.

Experimental program. The graph must be read from left to right in accordance with the timeline. (A) The actions performed by the participant during the motor task (top, A-low; middle, A-high; and bottom, L-low). (B) Example of wrist trajectory (solid line) obtained during three movements in its anatomical reference plane during the pre-fixation phase. The dotted line connects the joint centers of the shoulder (s), elbow (e), and wrist (w), captured at the beginning of the extension exercise, maximum flexion of the elbow, and the end of the extension exercise. (C) The fixed period (orange bar) and VR session (blue diamond) based on the actions or natural scenes of the AOT and CTRL groups, respectively. (D) The kinematics (wrist trajectory and joint center position) after fixation is the same theme as shown in B, drawn together with the trajectory before fixation (gray line). Note the flexion of the elbow joint after the fixation, resulting in a smaller wrist trajectory.

No significant differences were found between the baseline AOT and CTRL groups in terms of duration of reaching (RD), peak velocity of reaching (VP), or motion segmentation (MFr; all P> 0.18) (SI appendix, Table S1). Table 1 reports the RD, VP and MFr results of the three test actions. According to the analysis of variance, all variables show significant TIME main effects. The most serious injury was observed immediately after removing the bandage, that is, during the first post-fixation test (T1), all subjects experienced significant recovery during the post-fixation test (from T1 to T10). This finding suggests that 16 hours of immobilization is sufficient to cause subtle but quantifiable changes in kinematic performance, the recovery of which can be evaluated in the post-immobilization procedure.

Differences between T1, T4, and T9 scores after fixation and the average scores before fixation for RD, VP, and MFr

The most interesting finding is indeed related to the influence of the GROUP factor, which seems to be limited to MFr, which is a parameter that indexes the relative ratio between the range of motion (ROM) of the elbow and shoulder angles of interest. Comparing the scores before and after fixation, the Mfr scores of the CTRL participants were higher than the AOT group of all three exercises [F(1, 38) = 7.55, P = 0.009, part η2 = 0.16, Bayes factor (BF) 10 = 5.15]. MOVEMENT also showed a significant main effect, which may reflect the stronger influence of the fixation procedure on the coordination of the shoulder and elbow joints in L-low motion. However, no interaction was found, proving that AOT has similar effects on the three movement modes. More detailed statistical analysis results of RD, VP and MFr are reported in Table S2 of the SI Appendix.

Since MFr depends on the angle of the shoulder and elbow joints, we separately evaluated the effects of fixation (time factor) and AOT (group factor) on each joint. The linear fitting method [LFM (38)] is used to compare each post-fixation test with the average pre-fixation kinematics. The method returns three indicators (amplitude modulation [AM], R2, amplitude offset between curves [AOff ]) Describe the movement patterns of different characteristics. Considering that the MFr score varies according to the ROM of the elbow and the shoulder, we pay attention to the AM parameter, which indicates whether the amplitude of the motion after fixation (AM> 1 and AM <1 respectively) is better than the baseline (T0). Larger/smaller. On the contrary, we did not anticipate the changes in the R2 and AOff indicators, which tested the difference in the time course and offset of the joint angle curve respectively.

Figure 2 depicts the average time course of the elbow flexion and extension angles in the first test of the AOT and CTRL groups in the pre-fixation and post-fixation stages. The results showed that fixation reduced elbow ROM, especially for the CTRL group. To support this, AM's analysis of variance (Figure 2, right) shows a significant main effect of GROUP [F(1, 38) = 10.98, P = 0.002, part η2 = 0.22, BF10 =15.94], using CTRL to participate The participants systematically showed lower AM scores: that is, compared to the AOT group, all exercises had higher joint stiffness.

(Left and middle) The time course of the average elbow angle of the AOT and CTRL groups (green and red respectively) of the three movements (A-Low, A-High and L-Low) in the pre-fixation test (light color); The time course of the average elbow angle obtained in the first test in the post-fixation phase (after T1) is superimposed in the corresponding dark color. (Right) Evaluate the average AM and SE of the AOT and CTRL groups (green and red, respectively) at T1, T4, and T9.

Discover the main influence of time [F(2, 76) = 10.06, P <0.01, part η2 = 0.21, BF10> 100]. Post-hoc comparisons showed that the value of T1 was systematically lower than T4 and T9 (all P <0.05) (SI appendix, Table S2), emphasizing that the initial contraction of elbow movement was reversed during post-fixation training. There is no significant GROUP × TIME interaction, and a significant TIME effect indicates that the recovery dynamics of all exercises are not affected by AOT. On the contrary, AOT seems to retain the participants' original exercise ability, reflecting that the degree of injury at T1 is lower than that of CTRL, which persists throughout the postfixation period.

No significant influence of the factor GROUP on shoulder movement was observed, but there was a significant influence of TIME [F(2, 76) = 9.05, P <0.001, part η2 = 0.19, BF10> 100]. However, it should be noted that the AM value of the shoulder is usually higher than the value of the elbow, which indicates that the shoulder angle is not affected by fixation (SI appendix, Figure S1 and Table S2). This difference may be due to the different degree of restraint imposed by the bandage on the two joints. Although the elbow is restricted to a fixed position, there is no remaining flexion or extension opportunity, but the shoulder retains some remaining mobility, which may obscure the overall impact of the fixation on the kinematics of the shoulder. The factor analysis of R2 and AOff shows that there is no shift between time patterns or curves that fixation does not affect the kinematics of the shoulder or elbow (SI Appendix, Table S3). The insignificant effect on R2 and AOff excludes any deviation in determining the difference in AM values.

Given that the distance between the participant and the object remains constant throughout the experiment, it is reasonable to ask whether the reduction in elbow flexion and extension can be attributed to the different dynamic posture adjustments of the torso before and after the fixation. However, we ruled out this possibility by calculating the displacement of the torso during exercise and verifying that there is no significant GROUP effect (SI Appendix, Table S3).

In this study, sexually transmitted infections caused changes in sports performance, and participants showed longer exercise duration, higher MFr, and smaller range of motion after immobilization. These findings are consistent with previous evidence that STI impairs the performance of restricted body parts, even after a fixed period of 10 to 12 hours (33, 39, 40). The changes observed in the two groups appeared to be largely reversible, with exercise performance almost completely restored in 10 trials after immobilization (Reference 40 has similar results).

The STI of healthy participants provides a neurobehavioral model to explore the efficacy of implicit motor interventions (for example, motion observation or motor imagination), which can adaptively stimulate cortical motor representations in the context of maladaptive neuroplasticity. It is not affected by disease-related confounding factors. 36, 41⇓ –43). One might argue that the young age of our study participants (average age 22.5 years) limits the generality of our findings to the elderly population. However, previous studies have shown that when comparing older people with younger people, the activities of the fronto-apical network shared by action observation, motor imagination, and action execution (44) did not show any age-related changes (45), thus supporting us The universality of the discovery of different age groups.

Most interestingly, our data showed that participants who received AOT during the fixed period had better athletic performance at the end of the fixed period (Figure 3). It is worth noting that this effect is related to the spatial organization of the movement, which is reflected in the higher MFr scores of the CTRL group. On the contrary, the temporal characteristics of exercise are less affected by AOT intervention. This difference may be due to the neural basis of movement observation, which relies heavily on the frontal-parietal network that encodes motor tissue (1), and only to a lesser extent on the neural basis for movement time organization (46⇓ –48).

Use AOT as a tool to prevent movement disorders caused by non-use of limbs. The continuous gray line represents the hypothetical time course of limb motor ability before the start of immobilization. Athletic capacity is reduced when injured (i.e. starting to brake) and is indicated by a dashed line. The red and green solid lines that start after the offset is fixed indicate the recovery of motion in the CTRL and AOT groups, respectively. "AOT benefit" refers to the advantages provided by AOT in terms of residual exercise capacity during the fixation process.

Which neural mechanism enables AOT to prevent the decline in exercise capacity of healthy volunteers undergoing STI? It is well known that sexually transmitted infections can cause cortical motor depression characterized by nerves in fixed limbs (41, 49, 50). In the long run, this reduced excitability may promote the emergence of maladaptive behaviors or the consolidation of compensatory attitudes. Although initially beneficial to the patient, it is usually harmful to the long-term outcome (51). For these reasons, the development of early-onset interventions against cortical motor inhibition may play an important role in limiting the gradual deterioration of athletic performance.

In this regard, AOT has been shown to effectively limit STI-induced reorganization of cortical maps. Basolino et al. (41) It was proved that after 10 hours of upper limb immobilization, healthy volunteers who received motion observation stimulation almost completely retained the cortical motion picture, while the corticospinal excitability of CTRL participants who did not receive AOT was greatly reduced. These findings indicate that AOT has the ability to counteract the inhibition of cortical movement after limbs are not used, and they may explain the neural mechanism of maintaining sports performance in the AOT group in our study. In view of our experimental design, whether the fixed reduction effect is mainly driven by observing a particular movement, or rather, observing any movement is still an open point. However, the viewpoint that motion observation triggers motor activity after body position and muscle position organization (52) points out that consistent motion is the ideal stimulus for AOT. This is also consistent with the previous behavioral data, indicating that during physical exercise tasks, movement observation can lead to the enhancement of motor memory coding, but the premise is that the observed movement is consistent with the exercised movement (53).

Although our research involves healthy subjects, the results have reference value for different clinical situations. Orthopedics and peripheral nervous system diseases (for example, brachial plexus neuropathy, nerve or nerve root damage, Guillain-Barré syndrome and other inflammatory diseases) represent the clinical conditions closest to our experimental model, because the brain that hosts the mirroring mechanism The structure is complete. In all these cases, cortical motor inhibition may lead to the instantiation of dysfunctional motor behaviors. AOT can promote the maintenance of the central-peripheral interaction similar to that before the onset, which is beneficial to the faster recovery of motor function.

In central nervous system diseases (such as stroke), the patient’s inability to move is not due to peripheral restrictions, but due to damage to the brain structure responsible for generating and controlling movement. After a sudden interruption of a motor program, the motor system undergoes compensatory neurological processes, such as remapping around the lesion (54, 55) and rebalancing of functions between hemispheres (56, 57); therefore, the exclusive and progressive view of cortical motor inhibition is This is not appropriate. In this case, maladaptive neuroplasticity processes may occur, and AOT can still limit their instantiation. This benefit is consistent with the well-known effectiveness of AOT in promoting recovery from exercise in patients after stroke (4, 5). However, this ability depends to a large extent on the extent and topography of the lesion: that is, on the post-injury function of the cortical motor system. A promising aspect of our findings regarding post-stroke patients is the ability of AOT to primarily intervene in MFr, which has been described as a key feature of stroke-related motor dysfunction (58, 59).

This study shows that the use of AOT during immobilization limits the movement changes caused by the non-use of the limb. In view of the protective effect of AOT on the decline of exercise capacity, motion observation represents an effective tool for early intervention of the motor system during the period of non-use of the limbs, thereby reducing the burden of further rehabilitation.

Forty naive volunteers (17 men and 23 women, average age 22.9 ± 3.7) participated in the experiment. The sample size is calculated using power analysis, the acceptable minimum significance level (α) is 0.05, the acceptable power level (1 − β) is 0.80, and the effect size is 0.5 (part η2 = 0.2); use G* Power 3.1 (60) for analysis. Among the participants, 37 were right-handed and 3 were left-handed [according to the Edinburgh Dominant Handed List (61)]. All subjects reported normal vision or normal vision after correction, no history of neurological disease or recent orthopedic injury of superior upper limbs. Volunteers were randomly assigned to two experimental groups: 1) AOT: 9 men and 11 women, with an average age of 22.5±2.6 and 2) CTRL: 8 men and 12 women, with an average age of 23.4±4.6.

The local ethics committee approved the study (Comitato Etico dell' Area Vasta Emilia Nord, 10084, 12.03.2018), which was conducted in accordance with the principles stated in the Declaration of Helsinki. Each participant provided written informed consent before the experiment.

The main upper limbs of each participant were immobilized for a total of 16 hours. The arms and forearms are wrapped with orthopedic bandages commonly used in clinical practice. The bandage restricts the movement of the arms and shoulders by fixing the elbow joint in a 90° flexion (Figure 1C). The subjects were instructed to keep the bandage (even at night) and minimize upper limb movement from the time of bandage use (approximately 6:00 pm) to the next morning (approximately 10:00 am).

We chose a 16-hour fixation period because it imposes a period of non-use that may be sufficient to cause observable effects; in fact, previous studies have reported that the upper limbs after fixation can be observed 10 to 12 hours after the start of fixation Significant behavioral changes in sports (34, 39, 40).

After removing the bandage, the participant immediately underwent a kinesiology assessment. They were instructed to maintain a relaxed posture and avoid any movement until the movement task began.

Because the study evaluated the impact of AOT on maintaining motor function during immobilization, the participants’ exercise performance was tested before (ie before immobilization) and after (ie, after immobilization) the upper limbs were not in use.

In the pre-fixation phase, the experimental device requires participants to perform three actions, which require reaching and grasping spheres (7 cm in diameter) placed in different positions in the surrounding space. The sphere is placed on the boundary of each participant's personal space: that is, at one arm's distance, thereby limiting the torso tilt during task execution.

Participants sit comfortably on a stool while maintaining a neutral starting position with their hands on their knees. They were instructed to perform the following upper limb movements (Figure 1A):

A) Reach and grasp the sphere at A-Low;

B) Reach and grasp the sphere at the height of A;

C) Reach and grasp the sphere at L-Low, which fits the participant's dominant hand.

After each action is completed, the participant is asked to return to the starting position. The sequence of the three actions (ABC) is repeated 10 times, for a total of 30 actions.

After the kinesiology assessment, the participant was immobilized as described above. Three virtual reality (VR) therapy sessions (with "action" or "natural scene" stimulation) were conducted for each participant. The subjects received the first treatment immediately after immobilization, and the next morning, just before the bandage was removed, the other two treatments (20 minutes apart). At the end of the VR course, the bandage was removed and the participant immediately performed the same exercise task as the pre-fixation phase (Figure 1D).

The visual stimulus was developed using immersive VR technology using a cross-platform game engine (Unity 3D; Unity Technologies) and presented to participants through a VR headset display (HTC Vive Pro; HTC Corporation).

Two types of stimuli were created; participants in the AOT group observed the "hands-out" task from a first-person egocentric perspective, while participants in the CTRL group observed three different scenes, which depicted no movement The natural scene of the content. In the AOT scenario, the humanoid avatar performs the same sequence of actions as the participants are required to perform, thereby mimicking the experimental environment. The avatar is animated using the kinematics obtained from the actor.

When wearing a VR headset, subjects can rotate their heads in a virtual 3D (3D) environment until the movement appears in their field of view. More precisely, the subjects looked forward during the A-Low and A-High movements, and they rotated their heads 90° (clockwise or counterclockwise, depending on the subject’s dominant hand) to perform L- Low sports.

The subject observes the stimuli, and each stimulus lasts about 1 minute. In the case of an action, the virtual stimulus consists of a single action repeated 10 times. After observation, the participant performs a motor imagery task: that is, closing the eyes and at the same time performing a 30-second mental rehearsal of the action just observed. For each action (3), this 90-second block is repeated four times, with a total duration of approximately 18 minutes. The course arrangement of the CTRL group remains unchanged, replacing action stimuli with natural scene stimuli, and replacing motion imagery with visual memories of natural scenes.

Participants' movements are recorded using a marker-based 3D photoelectric system (SMART; BTS Bioengineering), which consists of six infrared cameras that detect the position space of eight reflective spherical markers (10 mm in diameter) at a sampling rate of 120 Hz The resolution is 0.3 mm. According to the recommendations of the International Society of Upper Limb Biomechanics (62), place the marker on the subject's dominant arm: one mark on the suprasternal notch; one mark on the seventh cervical vertebra; one mark on the spinous process of the eight thoracic vertebra; There is a mark on the shoulder at the edge of the acromion; two marks on the lateral and medial epicondylar elbow of the humerus; two marks on the radial styloid and ulnar styloid of the wrist. After reconstructing the positions of the markers, their original trajectories were post-processed (Matlab2018a; MathWorks Inc.), including low-pass filtering and reconstruction of missing marker positions (63).

Motion analysis is focused on the arrival stage, segmented according to the tangential velocity of the ulnar styloid mark (64). The time boundary of the arrival phase was determined as the time when the ulnar velocity exceeded and returned below 5% of the peak velocity (respectively reaching the start and end points). Once the arrival stage is defined, specific parameters are calculated to evaluate the subject's performance (65):

• RD: the total duration of arrival movement (64);

• VP: the peak value of the tangential velocity at the ulnar styloid mark (66);

• MFr: An index that quantifies the relative ratio between the range of motion of the elbow and shoulder, used to assess whether the overall kinematics method is different before and after fixation and between groups; this index is based on the report in (67) The method is defined as: MFr=100%(1−EROMSROM), where EROM is the average ROM elbow flexion-extension, and SROM is the average ROM of the shoulder angle (shoulder flexion-extension) or abduction-adduction). A decrease in elbow joint ROM will lead to a higher MFr score, while a decrease in shoulder joint ROM will lead to a lower MFr score.

Subsequently, the shoulder joint angle was calculated according to the recommendations of the International Association of Biomechanics for the upper limbs (62). For A-Low and A-High exercises, shoulder flexion and extension (sagittal plane) are considered, and for L-Low exercises, shoulder abduction-adduction (coronal plane) is considered. Flexion and extension are considered the most important for the elbow joint. These angles are hereinafter referred to as angles of interest.

In order to make the movement time patterns within and between participants comparable, all kinematic angles are segmented and resampled to a standardized time interval of 0 to 100%, thereby scaling all movements to a standard duration. These data can then be compared based on movement patterns, regardless of their duration. LFM (38) is used to analyze the waveform similarity for each angle of interest; specifically, LFM compares each post-fixation test with the average pre-fixation performance. LFM returns three independent indices: 1) the linear goodness of fit (R2) between the curves, 2) AOff, and 3) the AM index. R2 ranges from zero to one (from zero to perfect similarity between the curves). The ideal AM value is 1, which represents the greatest similarity in amplitude between the curves (a value lower than 1 indicates that the angle under study is reduced; a value higher than 1 indicates an increase in amplitude). In view of our experimental design, the AM parameter is used to test the effect of fixation and the influence of AOT on the amplitude of the angle of interest, R2 is used to reveal whether the temporal motion tissue changes before and after fixation, and AOff is used to verify whether there is a measurement error. The last two indices are used as control variables to test the reliability of the AM results and exclude any deviation effects due to curve movement (68).

In addition, although the object to be stretched out and grasped is systematically located in the surrounding space of the subject, so there is no need for any inclination of the torso during exercise, but we verify the arrival process by calculating the maximum displacement of the incision mark on the sternum There is no compensatory trunk movement from its initial position.

All in all, we obtained 10 performance parameters and LFM coefficient values ​​for each action and theme. Calculate the within-subject difference of RD, VP, and MFr relative to baseline. These variables represent the trial-to-trial changes in exercise performance over time (T1–T2–T3–T4–T5–T6–T7–T8–T9–T10) relative to the average baseline value (T0) after fixation.

A series of unpaired t-tests were performed on the pre-fixed scores of each kinematic parameter between groups to exclude the possibility of any significant difference between AOT and CTRL at baseline (SI Appendix, Table S1).

A mixed design analysis of variance was performed on the main parameters (RD, VP, MFr, AMElb, AMSho), with MOVEMENT (A-Low, A-High, L-Low) and TIME as the in-subject factors and GROUP (ie, AOT/ CTRL) as the inter-subject factor. The same statistical design is applied to the control parameters (R2Elb, AOffElb, R2Sho, AOffSho, Trunk) to exclude the possibility of any significant influence of TIME and GROUP factors.

Previous literature has shown that functional recovery in rehabilitation usually follows a logarithmic trend (69⇓ –71). We use this principle to reduce the number of levels within the subject. In other words, we have identified the tests that are estimated to achieve 50% and 95% recovery rates: T4 and T9, respectively. Therefore, the trials of T1, T4, and T9 were used as in-subject factors for analysis of variance.

Considering the number of analyses of variance, we applied the Šidák-Bonferroni correction to the main analysis and set the significance threshold to P = 0.01 (0.05 divided by five variables). Use Newman-Keuls correction to perform post-hoc testing for multiple comparisons. All variables are normally distributed, verified by Kolmogorov-Smirnov test (P> 0.05). The part η2 is calculated as a measure of the size of the effect.

Implement Bayesian statistics to measure the probability of the true influence of test factors on kinematics. Calculate BF to indicate how many times the alternative hypothesis (H1) of the primary variable (BF10) is more likely to occur than the null hypothesis (H0), or how many times H0 is more likely to occur than the control variable H1 (BF01). The BF robustness check (SI appendix, Table S2) indicates the level of evidence (72). 

The original data used in this study can be obtained at the following link of Zenodo: https://doi.org/10.5281/zenodo.5603250.

This research was supported by the "Kinect-Hololens Assisted Rehabilitation-KHARE" project, which was funded to IN-CNR by INAIL (Istituto Nazionale per l'Assicurazione contro gli Infortuni sul Lavoro) and Fondazione Cariparma.

↵1D.DM, ES, MF-D. and PA have made the same contribution to this work.

Author contributions: DDM, ES, ET, MF-D. and PA design research; DDM, ES and MCB conducted research; DDM, ES, MCB, AN, NFL and PA analysis data; and DDM, ES, AN, GR , MF-D. and PA wrote this paper.

Reviewers: FB, Rheinisch-Westfalische Technische Hochschule Aachen; GK, Central European University, Budapest; and SS, University of Texas at Dallas.

The author declares no competing interests.

This article contains online support information https://www.pnas.org/lookup/suppl/doi:10.1073/pnas.2025979118/-/DCSupplemental.

This open access article is distributed under Creative Commons Attribution 4.0 (CC BY).

Thank you for your interest in advertising on PNAS.

Note: We only ask you to provide your email address so that the people you recommend the page to know that you want them to see it, and that it is not spam. We do not capture any email addresses.

Feedback privacy/legal

Copyright © 2021 National Academy of Sciences. Online ISSN 1091-6490. PNAS is a partner of CHORUS, COPE, CrossRef, ORCID and Research4Life.