Efficacy and safety of using auditory-motor entrainment to improve walking after stroke: A multi-site randomized controlled trial of the InTandemTM neurorehabilitation system

doi:10.21203/rs.3.rs-3031294/v1

Download PDF

Article

Efficacy and safety of using auditory-motor entrainment to improve walking after stroke: A multi-site randomized controlled trial of the InTandem^TMneurorehabilitation system

https://doi.org/10.21203/rs.3.rs-3031294/v1

This work is licensed under a CC BY 4.0 License

Journal Publication

published 08 Feb, 2024

Read the published version in Nature Communications →

Version 1

posted

You are reading this latest preprint version

Background: Walking slowly in the chronic phase of stroke limits community ambulation and reduces long-term health and quality of life. Our objective was to evaluate the safety and efficacy of an autonomous neurorehabilitation system designed to utilize auditory-motor entrainment to improve walking in people in the chronic phase of stroke recovery.

Methods: The multi-site, prospective, interventional, 2-arm randomized controlled trial enrolled 87 individuals >6 months post-stroke and with walking speeds between 0.5 m/s and 0.8 m/s. Participants were asked to complete 15 walking intervention sessions over a 5-week period and were assigned to either experimental (i.e., walking with the MR-001 neurorehabilitation system [InTandem^TM]) or Active Control (i.e., walking without InTandem) interventions. The primary endpoint was the between-group difference in the change in walking speed, as measured by the 10-meter walk test (10mWT) before and after each intervention period. Between-group differences in the rate of responders and time course of walking speed improvement were also assessed. Responders were those with a: (1) speed increase >0.16 m/s or (2) speed increase >0.16 m/s and a post-training speed above 0.80 m/s. Safety was assessed by the frequency of adverse events (AEs).

Results: InTandem resulted in a larger increase in speed (Δ: 0.14±0.03 m/s) compared to Active Control (Δ: 0.06±0.02 m/s) (F(1,49)=6.58, p=0.013). Participants randomized to InTandem were over 3x more likely to be responders (χ²(1)≥6.47, p=0.01). InTandem was also associated with a 2.7x faster rate of speed increase over the 15-session intervention period (F(16,206)=1.790, p=0.034). Both groups experienced the same number of AEs and serious AEs.

Conclusions: InTandem uses the closed-loop control of music to autonomously individualize and progress walking rehabilitation. While further study is warranted to better understand the ideal treatment length, durability of effect, and healthcare resource utilization impact, InTandem provides a safe and effective intervention that improves walking speed in the chronic phase of stroke.

Trial registration: NCT04121754

Health sciences/Neurology/Neurological disorders/Stroke

Health sciences/Medical research/Translational research

auditory-motor entrainment

rhythmic auditory stimulation

neurorehabilitation

walking

gait

walking speed

music

Disability after stroke is an important public health issue. The dearth of effective rehabilitation therapies¹ has resulted in stroke remaining a leading cause of long-term disability.^2–4 According to the American Heart Association, the medical-related costs of stroke are estimated to be $50 billion and are projected to more than double by 2035.⁴ The personal and societal burdens of stroke necessitate addressing existing care gaps.

A key care gap is the limited efficacy of walking rehabilitation in the chronic phase of stroke recovery. People with chronic stroke impairment identify walking difficulty as a major problem⁵ and regard community ambulation “as either essential or very important” to their well-being.⁶ Community ambulation is defined as “independent mobility outside the home” and includes confidently navigating uneven terrain, private venues, shopping centers, and other public settings.⁷ However, as many as 85% of individuals with chronic stroke impairment have at least some limitation in community ambulation.⁸ To navigate community settings successfully, a minimum walking speed of 0.80 m/s⁹ is often recommended. Whereas the average walking speed of community-dwelling older adults is 1.3 m/s¹⁰, individuals in the chronic phase of stroke have an average speed of only 0.74 m/s.⁶ That is, the average patient likely does not walk fast enough to navigate most community settings.

Slow walking is associated with a high fall risk, increased comorbidities, and reduced quality of life.^11–15 Interventions that are effective in improving the walking speed of people with chronic hemiparesis after stroke are critically needed; however, an additional care gap that exists in parallel with, and contributes to the limited efficacy of post-stroke walking rehabilitation is limited access to skilled interventions. Disparities in time, childcare, work responsibilities, and access to reliable transportation¹⁶ often prevent community-dwelling individuals with mobility impairment from engaging in the rehabilitation needed for long-term health.⁹ Together, care gaps in treatment efficacy and access necessitate the advancement of interventions that can provide safe and effective walking rehabilitation across care settings, including the home and the community.

We present the InTandem autonomous neurorehabilitation system as a promising solution to address these care gaps. InTandem aims to improve the walking of individuals in the chronic phase of stroke recovery by autonomously applying the principles of rhythmic auditory stimulation.^17–28 The system’s primary mechanism of action is auditory-motor entrainment, a neurally-mediated process whereby the timing of motor movements is involuntarily synchronized with the timing of a rhythmic auditory stimulus (e.g., music featuring a strong beat). The involuntary synchronization between movement and auditory rhythm allows InTandem to autonomously provide a progressive intervention by adjusting elements of the auditory rhythm, like tempo and beat salience. Notably, the rhythm-based intervention is individualized to each user’s unique gait pattern by the combination of real-time gait sensing with closed-loop music control algorithms, eliminating the need for clinician and user input to safely increase walking speed within and across intervention sessions.

Two proof-of-concept studies conducted with InTandem research prototypes demonstrate its clinical promise.^29,30 The first proof-of-concept study reported clinically-meaningful gains in walking speed after one intervention session.²⁹ Additional sessions resulted in continued improvements in walking speed, with an average increase of 0.12±0.03 m/s observed after three sessions. The second proof-of-concept study reported a 22±10% median reduction in multiple measures of temporal gait asymmetry and a median 9±5% reduction in the energetic cost of walking after one session.³⁰ These findings are indicative of improvements in the neuromuscular control of walking, which is important given that the gait of individuals post-stroke is not only slow, but also highly asymmetric and effortful³¹—factors associated with reduced physical activity and worse long-term health.³²

Building on this foundational prior work, we conducted a multi-site, prospective, interventional, 2-arm randomized controlled trial to evaluate the safety and efficacy of InTandem relative to a treatment-matched Active Control group. Importantly, though the primary outcome of this study of the autonomous InTandem intervention is an improvement in walking speed, InTandem’s algorithm incorporates real-time assessment of gait quality to determine when progression is appropriate. That is to say, InTandem balances walking safely alongside walking faster; this feature of the intervention is believed to be a critical element to autonomous use of InTandem in free-living settings by individuals at high risk of falling, such as individuals with chronic post-stroke hemiparesis. Safety was evaluated as the frequency of adverse events (AEs). Efficacy was evaluated by testing the hypothesis that, compared to the Active Control group, InTandem would result in a greater increase in walking speed and a greater percentage of responders. We also tested the exploratory hypothesis that the individualized and progressive intervention would result in a faster rate of walking speed improvement over the 5-week intervention period.

The multi-site, prospective, interventional, 2-arm randomized controlled trial (NCT04121754) of the safety and efficacy of the InTandem neurorehabilitation system enrolled 87 participants between September 2019 and January 2022. The trial was approved by the Institutional Review Boards for each participating center as follows: Advarra for Carolinas Medical Center, Boston University Charles River Campus IRB (which oversaw activities at both Boston University and Spaulding Rehabilitation Hospital), IRB of the Icahn School of Medicine at Mount Sinai, Johns Hopkins Medicine IRB, Kessler Foundation IRB, Northwestern University IRB, and UNC at Chapel Hill Non-Biomedical IRB. Written informed consent was secured for all participants. All procedures were in accordance with institutional guidelines.

Inclusion and exclusion criteria

Inclusion criteria for the trial consisted of the following: ≥50 years of age, greater than six months post-stroke, a baseline speed ≥0.50 m/s and <0.80 m/s, gait asymmetry observed by the study investigator, able and willing to consent, no cognitive impairment as evidenced in score ≤1 on question 1b and a 0 on question 1c on the NIH Stroke Scale, and ability to safely participate in protocol-defined sessions of 30-minute duration. Exclusion criteria for the trial consisted of the following: participating in another trial to evaluate an investigational drug at the time of enrollment, previous completion of the trial, unable or unwilling to provide informed consent, unable to participate in protocol-defined sessions without the use of assistive devices (e.g., a cane or walker), a baseline walking speed >0.80 m/s, known history of neurologic (excluding stroke) injury, more than 2 falls in the previous month, active participation in another walking rehabilitation intervention (e.g., physical therapy), use of an external lower limb prosthetic (i.e., artificial limb), hearing impairment that prevents interaction with the InTandem system, self-report of orthopedic surgery in the last year, severe aphasia and/or a speech/language disorder that limits the ability to express needs and comprehend instructions, comorbidities that prevent participation in a rehabilitative walking intervention, or vulnerable populations deemed inappropriate for participation in the trial.

Data Collection Overview

Trial data were collected at a baseline screening assessment, during each intervention session, and at a trial closing visit. The baseline screening consisted of administration of NIH Stroke Scale question 1b and 1c and the collection of basic demographic and medical history data. Medical history data were reviewed throughout the trial to identify changes. The 10-meter walk test (10mWT; comfortable speed) was used to measure the trial’s primary endpoint and was conducted at the baseline screening, the trial closing visit, and to start and end each intervention session. AEs were collected at each trial visit, or volunteered by the participant between trial visits, and captured on a standard form. The standard AE form collected a description of the event, the onset and resolution dates (or if the event was ongoing), the severity, management/treatment, outcome, and determination of the relationship to the intervention.

Primary Endpoint Data Collection

The trial’s primary endpoint was a between-group difference in the change in self-selected comfortable walking speed, as measured using the 10mWT (i.e., post-intervention 10mWT speed – pre-intervention 10mWT speed). The 10mWT is a timed walk test on a 10-meter walkway, with only the middle 6 meters timed. The 10mWT speed is the average walking speed from three separate 10mWT trials and is computed in meters per second (m/s). Assistive devices (i.e., cane, crutch, walker, and/or functional electrical stimulation device) were not allowed during 10mWT assessments; however, lower limb orthoses and braces were permitted if necessary for safety. Participants were blinded to the results of their 10mWT.

The primary endpoint analysis used a pre-intervention 10mWT speed collected no more than 14 days before the first session. If scheduling challenges required more than 14 days between the baseline assessment and the first session, the 10mWT speed measured immediately prior to the start of the first session was used. The post-intervention 10mWT speed used in the primary endpoint analysis was collected no more than 4 days after the final session. If scheduling challenges required more than 4 days between the trial closing visit and the final session, the 10mWT speed measured immediately prior to the final session was used. If participants did not complete the full 5-week intervention schedule or did not return to complete the trial closing visit, the 10mWT speed used in the intent-to-treat analysis was the 10mWT speed measured immediately prior to the final session completed.

Randomization

Eligible participants were randomized to one of two intervention arms: InTandem or a treatment-matched Active Control. Randomization was performed through the Electronic Data Capture (EDC) system (Castor, New York, New York) without stratification using block randomization with randomly selected block sizes to ensure approximately equal group sizes. Trial investigators and participants were not blinded to randomization. InTandem was supplied to trial sites by MedRhythms, Inc. as the MR–001 neurorehabilitation system (Portland, ME).

Trial Design Overview

This randomized controlled trial was conducted in clinical settings under the supervision of clinical investigators to evaluate safety and efficacy relative to a treatment-matched Active Control group. Because InTandem is designed for independent use in the home, a separate study has been designed to evaluate its independent set-up and operation (manuscript In Press ³⁶).

Regardless of intervention arm assignment, all participants were scheduled to complete 15 intervention sessions over a 5-week period, at a frequency of 3 sessions per week, and with each session consisting of 30 minutes of walking. A complete session was defined as 15-to–30 minutes of walking. Participants were not allowed to use assistive devices during the intervention session but were allowed to take rest breaks as needed or when investigators believed that a rest break was necessary. Because sessions ended automatically after 30 minutes, rest breaks counted towards each intervention session (i.e., a 5-minute rest break resulted in 25 minutes for intervention).

All intervention sessions occurred on an overground track or hallway of at least 100 feet in length. In addition, the space used for intervention sessions could not have any discernible background music or audible distractions that could compete with the auditory intervention. Though no cueing or walking instructions were provided to study participants during training, all study participants were provided general instructions at the start of each intervention session (see Appendix Afor the language recommended in the clinical trial protocol). During each intervention session, physiological parameters (i.e., heart rate, blood pressure, and/or respiration rate) were monitored according to each site’s standard procedures, or at an investigator’s discretion to ensure safety of each study participant. If a study participant’s measurements fell outside of the acceptable parameters set by the site or the study participant’s physician, the intervention session could be paused or terminated based on a collaborative decision made by the study participant and investigator.

If required, participants were allowed to restart their intervention schedule after a washout period (n = 6). The minimum washout period required was equal to the duration of intervention weeks already completed. Following the washout period, trial eligibility was re-evaluated. Only trial data collected after the washout period were used in the trial’s primary analyses.

InTandem System and Intervention

InTandem consists of a touchscreen device locked with a preloaded proprietary software application, a headset for delivery of the rhythmic auditory stimulus, two shoe-worn inertial sensors, and charging equipment (Figure 1). To produce the individualized and progressive intervention, personalized audio cues are embedded into time-shifted music based on real-time decisions made by closed-loop control algorithms that continuously assess the gait data collected by the inertial sensors.

InTandem’s closed-loop control of music includes two real-time algorithm components that operate in parallel. The first component assesses the user’s ability to entrain to the target tempo based on real-time measurements of step-to-beat alignment, defined as the ratio of the user’s walking cadence (i.e., steps per minute) to the tempo of the time-shifted music. Here “time-shifted” refers to both: (1) stabilization of the natural variability (if any) of the song’s original tempo and (2) the within-session adaptive tempo increases or decreases used to individualize the intervention. More specifically, the initial target tempo is set to the user’s baseline cadence, as measured by the inertial sensors during an un-cued baseline walk, with subsequent modulation of the target tempo based on assessments made by the second algorithm component during the intervention. The second component assesses the user’s gait symmetry and variability based on real-time sensor input during the intervention, with symmetry measured as a ratio of interlimb stance and swing times, and variability measured as the coefficient of variation of stride times.

The entrainment and gait quality values from the two components are then compared to proprietary thresholds, which the algorithm uses to make two decisions. The first decision made is if it’s appropriate to modulate the song tempo—doing so requires concurrence by both components that (1) the user is sufficiently entrained and (2) their gait quality is appropriate for progression; only if both criteria are achieved is it considered safe to increase the tempo. The specific criteria used to make these determinations are based on the developer’s clinical experience delivering rhythmic auditory stimulation interventions, which were further refined during development to accommodate walking in real-world settings.The second decision made is if the user requires the addition of a synchronized rhythm track to the music to enhance the beat salience. When the rhythm track is introduced, the volume of the music adjusts so that the user hears more of the rhythm track relative to the music. Once the user is entraining and their gait symmetry and variability is within acceptable ranges, the rhythm track is then removed. If a user does not entrain, even with the added rhythm track, the music of the tempo will decrease until the user is entraining. The cascade of decisions that tailor the intervention to a person’s gait on a given day allows for personalized treatment with every session.

The music used in the trial was screened to ensure therapeutic suitability. This process ensured that the music met requirements for beat prominence and tempo. More specifically, the music included with the InTandem system is a core part of the intervention; not every piece of music is fit for purpose. Using a proprietary screening process, we assess high and low level features of the audio content, such as average tempo (what cadences could this song work for?), time signature (is the song in duple or triple meter?), duration of song (is it too long or short of a song?), and beat strength (how prominent is the beat?), in addition to other feature analyses. Through this process, a song is either accepted or rejected for use in InTandem.

At the conclusion of each session, an automated voiceover instructed the user to walk for a brief period without music to allow for data collection without the auditory stimulus. No other instruction was provided at each session.

Active Control

The Active Control intervention was matched to the InTandem intervention in number, duration, and frequency of sessions. Like InTandem sessions, Active Control sessions consisted of 30 minutes of overground walking practice supervised by clinical investigators. Though Active Control participants did not wear the InTandem system, they did wear inertial sensors on their shoes to enable recording gait data during each session.

COVID–19 Considerations

The trial was disrupted by the COVID–19 PHE. In brief, an unplanned interim analysis was conducted during the research shutdown, requiring revision of the statistical plan. More specifically, alpha was adjusted from 0.05 to 0.025. Moreover, according to guidance from FDA, the impact of COVID–19 on the trial data was examined by assessing the comparability of participants enrolled before versus during the COVID–19 PHE. The investigation revealed an effect of COVID–19 that was reasonably explained by a non-random change in trial management at one center, and ultimately required administrative removal of 8 individuals to resolve the observed effect of COVID–19¹.

Statistical Analyses

Safety was described as the frequency, severity, and relatedness of treatment-emergent AEs. An AE was defined as any untoward or unfavorable physical or psychological occurrence and included any abnormal sign, symptom, or disease temporally associated with participation in the research. Descriptive statistics are reported using means and standard deviations (SD) for continuous variables and proportions for categorical variables, unless noted otherwise. A 2 x 2 General Linear Mixed Model (GLMM) was used to evaluate differences in the treatment effect between groups (i.e., a Treatment x Time interaction), which allowed for nesting participants within Centers. Effects of interest in the model included Treatment (InTandem vs. Active Control), Time (Pre-intervention vs. Post-intervention), and the Treatment x Time interaction. Box-Cox and Shapiro-Wilk tests were used to evaluate linearity and normality. Extreme outliers and influential cases were screened for and removed.³⁵ Robust standard errors were used to account for violations of normality. A compound symmetric covariance matrix was used to model the correlation structure among residuals and Satterthwaite Approximation was used to adjust the degrees of freedom for significance testing.

For the study primary endpoint of a change in gait speed overtime between groups, this study was originally powered (power = 0.80) to detect a moderately large to large effect (d = 0.64 to 0.82) for a plausible range of Intraclass correlation coefficients (ICCs), (r = 0.05 to r = 0.20), and conservative estimates of the correlation among repeated measures (r = 0.4 to r = 0.6) for n = 66 and alpha = 0.05.

Between-group differences in the number of responders were evaluated using Chi-Square (χ²) tests. Responders were defined using two criteria: (1) a post-intervention change in 10mWT speed that surpassed the 0.16 m/s minimal clinically importance difference (MCID)³¹ or (2) both a post-intervention change larger than 0.16 m/s and a final post-intervention speed above 0.80 m/s.³² To account for an unplanned interim trial analysis during the COVID–19 Public Health Emergency (PHE), alpha was lowered to 0.025.

In addition to the primary endpoint analyses, an exploratory time-course of change analysis was conducted using a 2 x 17 x 2 GLMM. This analysis tested between-group differences in the treatment effect using all 10mWT speed data collected. The model tested three main effects: Treatment (InTandem vs. Active Control), Time (Baseline, Session 1, …, Session 15, and Closing), and Trial (pre- vs. post- session walking speed), and their two-way interactions. Alpha was set to 0.05 for this analysis.

^[1]Eight individuals were found to have been recruited into the trial during the COVID-19 PHE shortly after first completing another walking intervention trial. The washout period imposed between the two competing trials was discovered to be inconsistent with the precedent set in the InTandem trial protocol for a washout period. This deviated recruitment strategy during the COVID-19 PHE was considered an unanticipated event that affected only these eight participants; their administrative removal resulted in COVID-19 no longer affecting the trial data.

Participants

A total of 87 individuals were enrolled and randomized. Of these, six withdrew after randomization and before completing their first session. The remaining 81 individuals completed at least one session and were thus to be included in the trial’s intent-to-treat analysis. However, of these 81, one extreme statistical and clinical outlier was identified and removed based on pre-planned outlier handling. This participant had a change in 10mWT speed of 0.79 m/s, which was 3.45 standard deviations beyond the group mean and would be considered extremely non-typical for five weeks of gait rehabilitation.³³ As noted in Methods: COVID–19 Considerations, eight additional participants were administratively removed from the trial after a COVID–19-related investigation into the trial data. Four of these study participants were from the InTandem intervention group and four were from the Active Control group.

Thetrial’s intent-to-treat analysis thus included 72 participants who completed at least one walking session. These study participants had an average (± standard deviation) time since stroke of 8.1 ± 7.1 years and were 62.3 ± 7.0 years of age. The cohort was 58.3% male, 43.1% Black or African American, and 80.6% Not Hispanic or Latino. Approximately 77% of participants completed some college or beyond. There were no significant demographic differences between groups (Table 1). Of these 72 participants, four randomized to InTandem and nine randomized to Active Control did not complete the full 15-session intervention schedule (Figure 2).

Session completion and InTandem system reliability

The trial included a total of 1,015 completed intervention sessions. Of these 1,015 sessions, 984 sessions (96.9%) were fully complete and 31 (3.1%) were terminated early. Of the 31 sessions terminated early, 15 were ≥15 minutes in duration and thus did not require restart or rescheduling as per the trial protocol. The remaining 16 early terminated sessions were <15 minutes and thus considered incomplete, requiring a session restart or rescheduling. Of these 16 incomplete sessions, 13 were incomplete due to software or system component issues. System-related issues were able to be rapidly resolved, allowing the sessions to be restarted and completed uneventfully on the same day without a need to reschedule. The remaining 3 incomplete sessions were incomplete due to the participant being too tired to continue (n = 1) and AEs (n = 2); none of the 3 sessions were able to be restarted on the same day. Two of these three incomplete sessions were rescheduled to a later date and completed successfully; the one remaining incomplete session was never rescheduled as the participant was withdrawn from the trial due to unrelated medical reasons.

Primary Endpoint Analyses

InTandem resulted in a larger increase in 10mWT speed compared to the Active Control intervention (F(1, 49) = 6.58, p = 0.013). Specifically, InTandem resulted in an average (± standard error) increase in 10mWT speed of 0.14 ± 0.03 m/s (p < 0.001) compared to the Active Control’s increase of 0.06 ± 0.02 m/s (p < 0.001) (Figure 3A). Moreover, InTandem resulted in a greater proportion of responders based on the trial’s two definitions of being a responder. Participants who received InTandem were 3.2x more likely to increase their 10mWT speed beyond the 0.16 m/s MCID (Active Control: 4/32, 12.5%; InTandem: 16/40, 40.0%; χ²(1) = 6.70, p = 0.01, Cramer’s V = 0.31) and 3.7x more likely to both increase their 10mWT speed beyond the 0.16 m/s MCID and have a post-intervention speed above 0.80 m/s (Active Control: 3/32, 9.4%; InTandem: 14/40, 35.0%; χ²(1) = 6.47, p = 0.01, Cramer’s V = 0.30) (Figure 3B).

Exploratory Time-Course of Change Analysis

The average 10mWT speed measured before and after each session changed by a different rate across the 5-week intervention period for the InTandem group versus the Active Control group (i.e., a significant Session x Treatment interaction, F(16, 206) = 1.790, p = 0.034). Though both the InTandem and the Active Control groups increased their 10mWT speed across sessions, individuals who received the InTandem intervention showed a markedly faster rate of walking speed increase across sessions (Figure 3C).

Safety: Frequency of adverse events

Seven InTandem users (17.5%) experienced a total of ten AEs and two serious adverse events (SAEs) over the course of the trial (Table 2). One SAE was deemed possibly related to InTandem because it occurred during a walking session; the event resolved within a day and the participant completed the trial on schedule. Similarly, in the Active Control group, seven participants (21.9%) experienced AEs, two of which were considered SAEs. Neither SAE was deemed related to the walking sessions.

There were six reported fall events during the trial: four falls in the Active Control group compared to two falls in the InTandem group. All six falls were non-injurious, occurred outside of trial visits, and were considered “not related” or “unlikely related.” There was no observed pattern to the timing of the fall events with respect to the 5-week training period; two falls occurred during study participants’ first week of training, and one fall occurred in each of the subsequent weeks of the training period (i.e., one in week 2, one in week 3, etc).

InTandem is an autonomous neurorehabilitation system that applies the closed-loop control of music to provide an individualized and progressive intervention tailored to each user’s unique gait pattern, without requiring input from clinicians or users.The findings of this multi-center randomized clinical trial demonstrate the safety and efficacy of InTandem for individuals with chronic walking impairment after stroke. When taken together with findings of high usability ³⁶, InTandem has potential to address key efficacy and access care gaps.

Faster walking speeds and rate of recovery

Autonomous rehabilitation delivered by the InTandem system resulted in significantly faster post-training walking speeds compared to the treatment-matched Active Control, with InTandem participants being three times more likely to be responders. Given the relationship between walking speed and health, function, and quality of life, success in achieving the trial’s primary endpoint of a between-group difference in the change in walking speed underscores the potential of the InTandem system to improve walking after stroke. Critically, InTandem participants also experienced a markedly faster rate of walking speed improvement throughout the five-week intervention period. Whereas the Active Control intervention required 16 trial sessions to produce an average walking speed change of 0.05 m/s, InTandem facilitated this change every 6 trial sessions.

The difference in the rate of speed recovery across the intervention groups has two major implications. On the one hand, during the early trial period (i.e., sessions 1 through 8), the effect of the two interventions on walking speed showed a rapid separation. On the other hand, during the latter half of the trial period (i.e., sessions 9 through 16), there was no evidence of a plateau in speed recovery in either group. This suggests that an even longer intervention period may produce even larger gains. Indeed, a recent randomized clinical trial by Boyne et al tracked changes in the walking speed of individuals with chronic stroke walking impairment that resulted from 12 weeks of high versus moderate intensity walking rehabilitation, reporting growing separation between the two intervention arms from 4 to 8 to 12 weeks of training.³⁴ These findings motivate further study of the effects of longer-term neurorehabilitation with InTandem.

Comparison of the findings of Boyne et al to our study suggests that InTandem may be more effective than both moderate-intensity aerobic training and high-intensity interval training. Indeed, whereas the moderate-intensity aerobic training group studied by Boyne et al performed similarly to our study’s Active Control group (i.e., both had a 0.06 m/s average increase in 10mWT speed), when compared to high-intensity interval training, InTandem resulted in larger gains in less time. More specifically, InTandem resulted in a 27% larger increase in 10mWT speed (i.e., 0.14 m/s vs. 0.11 m/s) in 37% less time (i.e., 5 weeks vs. 8 weeks). However, it should be noted that Boyne et al report continued gains in walking speed, up to 0.19 m/s, after 12 weeks of high-intensity training. Though our 5-week study is not able to directly compare against a 12-week intervention, our time-course of change analysis (Fig. 3C) predicts an over 0.30 m/s increase in walking speed after 12 weeks of InTandem intervention, which would be a noteworthy 58% larger increase in walking speed compared to high-intensity interval training. Though this prediction assumes a linear rate of recovery over 12 weeks, and certainly requires validation, InTandem’s potential to produce larger gains than high-intensity interval training in less time is clinically important and warrants strong consideration for patients who may not be good candidates for high-intensity exercise programs (e.g., due to comorbidities or tolerance).

Faster walking, but not at the expense of safety

InTandem was generally safe and well-tolerated, with an equal number of AEs and SAEs as reported for the Active Control group. Moreover, the number of treatment-emergent falls was fewer (i.e., two in the InTandem group versus four in the Active Control group). Critically, it should be noted that the similar number of adverse events was observed despite InTandem users walking markedly faster throughout most of the study period (Figure 3C). Though these results challenge the common belief that faster walking may be unsafe for individuals post-stroke, to address this reasonable concern, the autonomous InTandem intervention incorporates a multi-tiered decision-making framework to determine when progression of the music’s tempo is appropriate (see Methods).. The real-time assessment of gait quality included in this decision-making process is considered a key safety feature. By integrating continuous evaluations of gait quality into the system’s decision-making processes, future deployment of the InTandem intervention for targeted walking rehabilitation during everyday activities is likely to be safe and well-tolerated by individuals with chronic gait impairment after stroke.

Limitations

Several limitations to the trial must be acknowledged. First, the trial’s inclusion criteria limited the intervention to individuals with chronic post-stroke hemiparesis with walking speeds between 0.50 m/s and 0.80 m/s. Though this is the subgroup of people post-stroke we predicted would most benefit from this intervention, generalizability to other subgroups post-stroke, as well as other patient populations, is unknown. Second, the trial’s five-week intervention period was relatively short, and it is unknown if the observed rate of change would continue with longer treatment periods. Third, there was no follow-up period. In traditional rehabilitation RCT designs, follow-up assessments are often necessary to evaluate motor learning (i.e., retention) and the durability of the treatment effect; however, in contrast to traditional designs that evaluate the primary endpoint across three timepoints (i.e., pre-training, post-training, and follow-up), our time-course of change analysis allowed assessment of changes in walking speed across 17 timepoints (i.e., pre-training, each of the 15 training sessions, and post-training), providing evidence for marked motor learning (i.e., retention) session-over-session in the InTandem group that was largely absent in the Active Control group. However, without a follow-up assessment, the durability of InTandem’s effects are not known. Fourth, the trial did not include other important outcome measures, such as gait biomechanics, self-efficacy, and community walking activity; the full extent to which post-stroke walking can be improved by InTandem is thus not known. Finally, though study participants were blinded to their 10mWT speeds, the nature of the intervention prevented blinding of group assignment.

While the InTandem neurorehabilitation system is designed to provide an individualized and progressive intervention without requiring a clinician to provide real-time input, it is important to acknowledge that InTandem does not replace the need for a clinician prescriber. In practice, a clinician will identify appropriate candidates for the automated auditory-motor intervention and may be required to define patient-specific conditions for its implementation that go beyond the intervention itself—e.g., the use of adjunctive therapies and training aides, the level of supervision and guarding that may be required to walk, physiological parameter limits (e.g., maximum allowable heartrate), or appropriate settings of use.

The InTandem neurorehabilitation system is safe and effective in improving walking after stroke. Applying the closed-loop control of music to autonomously individualize and progress post-stroke walking rehabilitation has potential to address the unmet efficacy and access needs that currently limit recovery. While further study is warranted to better understand the feasibility and rehabilitative potential of using InTandem in home settings, the ideal length of treatment, the durability of effect, and healthcare resource utilization impact, the findings of this study support advancement of InTandem as a promising treatment option for individuals in the chronic phase of stroke.

Acknowledgements: We acknowledge contributions by the following study staff: at BU: Johanna Spangler, Lillian Braga-Ribeirinha, Joan Breen, Terry Ellis; at SR: Sara Prokup, Kristine Buchler, Matt Gifforn, Kelly McKenzie, Matt McGuire, Jen Traines; at KF: Melvin Mejia, Brandon Ross, Corey Greene, Kathleen Chervin, Sharon Franco, Greg Ames, Oluwaseun Ibironke; at UNC-Chapel Hill: Chelsea Duppen, Alex Huntsinger, David Rowland; at SRH: Catherine Adans-Dester, Anne O’Brien; at AH: Jillian Cummings; at JH: Kendra M. Cherry-Allen, Margaret A. French, Spencer Gonzaga.

Sources of funding: This trial was funded by MedRhythms, Inc.

Disclosures: LNA is a paid advisor to MedRhythms Inc. BH is cofounder and CEO of MedRhythms Inc with equity interest. ST and DP are employees of MedRhythms Inc with equity interest. This study was completed under a master management plan instituted by the Boston University Financial Conflicts of Interest Committee.

Data and materials availability: The data supporting the main conclusions of this manuscript are located within the main text and figures. Additional data are available upon request.

Advances and challenges in stroke rehabilitation - The Lancet Neurology [Internet]. [cited 2023 Apr 3];Available from: https://www.thelancet.com/journals/laneur/article/PIIS1474-4422(19)30415-6/fulltext
Ovbiagele B, Goldstein LB, Higashida RT, Howard VJ, Johnston SC, Khavjou OA, Lackland DT, Lichtman JH, Mohl S, Sacco RL, et al. Forecasting the future of stroke in the United States: a policy statement from the American Heart Association and American Stroke Association. Stroke. 2013;44:2361–2375.
GBD 2016 Stroke Collaborators. Global, regional, and national burden of stroke, 1990-2016: a systematic analysis for the Global Burden of Disease Study 2016. Lancet Neurol. 2019;18:439–458.
Virani SS, Alonso A, Benjamin EJ, Bittencourt MS, Callaway CW, Carson AP, Chamberlain AM, Chang AR, Cheng S, Delling FN, et al. Heart Disease and Stroke Statistics—2020 Update: A Report From the American Heart Association [Internet]. [cited 2023 Apr 6];Available from: https://www.ahajournals.org/doi/epub/10.1161/CIR.0000000000000757
Combs SA, Van Puymbroeck M, Altenburger PA, Miller KK, Dierks TA, Schmid AA. Is walking faster or walking farther more important to persons with chronic stroke? Disabil. Rehabil. 2013;35:860–867.
van de Port IG, Kwakkel G, Lindeman E. Community ambulation in patients with chronic stroke: how is it related to gait speed? J. Rehabil. Med. 2008;40:23–27.
Lord SE, McPherson K, McNaughton HK, Rochester L, Weatherall M. Community ambulation after stroke: how important and obtainable is it and what measures appear predictive? Arch. Phys. Med. Rehabil. 2004;85:234–239.
Fulk, G.D., He, Y., Boyne, P. and Dunning, K., 2017. Predicting home and community walking activity poststroke. Stroke, 48(2), pp.406-411.
Winstein CJ, Stein J, Arena R, Bates B, Cherney LR, Cramer SC, Deruyter F, Eng JJ, Fisher B, Harvey RL, et al. Guidelines for Adult Stroke Rehabilitation and Recovery: A Guideline for Healthcare Professionals From the American Heart Association/American Stroke Association. Stroke. 2016;47:e98–e169.
Ada L, Dean CM, Lindley R, Lloyd G. Improving community ambulation after stroke: the AMBULATE Trial. BMC Neurol. 2009;9:8.
Allali G, Launay CP, Blumen HM, Callisaya ML, De Cock A-M, Kressig RW, Srikanth V, Steinmetz J-P, Verghese J, Beauchet O, et al. Falls, Cognitive Impairment, and Gait Performance: Results From the GOOD Initiative. J. Am. Med. Dir. Assoc. 2017;18:335–340.
Lusardi MM, Fritz S, Middleton A, Allison L, Wingood M, Phillips E, Criss M, Verma S, Osborne J, Chui KK. Determining Risk of Falls in Community Dwelling Older Adults: A Systematic Review and Meta-analysis Using Posttest Probability. J. Geriatr. Phys. Ther. 2017;40:1–36.
Montero-Odasso M, Schapira M, Soriano ER, Varela M, Kaplan R, Camera LA, Mayorga LM. Gait velocity as a single predictor of adverse events in healthy seniors aged 75 years and older. J. Gerontol. A. Biol. Sci. Med. Sci. 2005;60:1304–1309.
Perera S, Patel KV, Rosano C, Rubin SM, Satterfield S, Harris T, Ensrud K, Orwoll E, Lee CG, Chandler JM, et al. Gait Speed Predicts Incident Disability: A Pooled Analysis. J. Gerontol. A. Biol. Sci. Med. Sci. 2016;71:63–71.
Studenski S, Perera S, Patel K, Rosano C, Faulkner K, Inzitari M, Brach J, Chandler J, Cawthon P, Connor EB, et al. Gait speed and survival in older adults. JAMA. 2011;305:50–58.
Ayala C. Use of Outpatient Rehabilitation Among Adult Stroke Survivors — 20 States and the District of Columbia, 2013, and Four States, 2015. MMWR Morb. Mortal. Wkly. Rep. [Internet]. 2018 [cited 2023 Apr 6];67. Available from: https://www.cdc.gov/mmwr/volumes/67/wr/mm6720a2.htm
Thaut MH, Kenyon GP, Schauer ML, McIntosh GC. The connection between rhythmicity and brain function. IEEE Eng. Med. Biol. Mag. 1999;18:101–108.
Grahn JA, Watson SL. Perspectives on Rhythm Processing In Motor Regions of the Brain. Music Ther. Perspect. 2013;31:25–30.
Kobinata N, Ueno M, Imanishi Y, Yoshikawa H. Immediate effects of rhythmic auditory stimulation on gait in stroke patients in relation to the lesion site. J. Phys. Ther. Sci. 2016;28:2441–2444.
Effects of Rhythmic Auditory Stimulation (RAS) on Gait Ability and Symmetry after Stroke [Internet]. [cited 2023 Apr 6];Available from: https://www.researchgate.net/publication/270440253_Effects_of_Rhythmic_Auditory_Stimulation_RAS_on_Gait_Ability_and_Symmetry_after_Stroke
Mainka S, Wissel J, Völler H, Evers S. The Use of Rhythmic Auditory Stimulation to Optimize Treadmill Training for Stroke Patients: A Randomized Controlled Trial. Front. Neurol. 2018;9:755.
Suh JH, Han SJ, Jeon SY, Kim HJ, Lee JE, Yoon TS, Chong HJ. Effect of rhythmic auditory stimulation on gait and balance in hemiplegic stroke patients. NeuroRehabilitation. 2014;34:193–199.
Thaut MH, Abiru M. Rhythmic Auditory Stimulation in Rehabilitation of Movement Disorders: A Review Of Current Research. Music Percept. 2010;27:263–269.
Crosby LD, Wong JS, Chen JL, Grahn J, Patterson KK. An Initial Investigation of the Responsiveness of Temporal Gait Asymmetry to Rhythmic Auditory Stimulation and the Relationship to Rhythm Ability Following Stroke. Front. Neurol. 2020;11:517028.
Cha Y, Kim Y, Hwang S, Chung Y. Intensive gait training with rhythmic auditory stimulation in individuals with chronic hemiparetic stroke: a pilot randomized controlled study. NeuroRehabilitation. 2014;35:681–688.
Lee S, Lee K, Song C. Gait Training with Bilateral Rhythmic Auditory Stimulation in Stroke Patients: A Randomized Controlled Trial. Brain Sci. [Internet]. 2018 [cited 2023 Apr 6];8. Available from: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6162464/
Ottawa Panel, Khadilkar A, Phillips K, Jean N, Lamothe C, Milne S, Sarnecka J. Ottawa panel evidence-based clinical practice guidelines for post-stroke rehabilitation. Top. Stroke Rehabil. 2006;13:1–269.
Hornby TG, Reisman DS, Ward IG, Scheets PL, Miller A, Haddad D, Fox EJ, Fritz NE, Hawkins K, Henderson CE, et al. Clinical Practice Guideline to Improve Locomotor Function Following Chronic Stroke, Incomplete Spinal Cord Injury, and Brain Injury. J. Neurol. Phys. Ther. 2020;44:49–100.
Hutchinson K, Sloutsky R, Collimore A, Adams B, Harris B, Ellis TD, Awad LN. A Music-Based Digital Therapeutic: Proof-of-Concept Automation of a Progressive and Individualized Rhythm-Based Walking Training Program After Stroke. Neurorehabil. Neural Repair. 2020;34:986–996.
Collimore A, Roto Cataldo A, Aiello A, Sloutsky R, Hutchinson K, Harris B, Ellis T, Awad L. Autonomous control of music to retrain walking after stroke. Neurorehabil. Neural Repair. In Press;
Tilson JK, Sullivan KJ, Cen SY, Rose DK, Koradia CH, Azen SP, Duncan PW, for the Locomotor Experience Applied Post Stroke (LEAPS) Investigative Team. Meaningful Gait Speed Improvement During the First 60 Days Poststroke: Minimal Clinically Important Difference. Phys. Ther. 2010;90:196–208.
Perry J, Garrett M, Gronley JK, Mulroy SJ. Classification of walking handicap in the stroke population. Stroke. 1995;26:982–989.
Boyne P, Miller A, Kubalak O, Mink C, Reisman DS, Fulk G. Moderate to Vigorous Intensity Locomotor Training After Stroke: A Systematic Review and Meta-Analysis of Mean Effects and Individual Response Variability [Internet]. Rehabilitation Medicine and Physical Therapy; 2022 [cited 2022 Nov 8]. Available from: http://medrxiv.org/lookup/doi/10.1101/2022.11.01.22281808
Boyne P, Billinger SA, Reisman DS, Awosika OO, Buckley S, Burson J, Carl D, DeLange M, Doren S, Earnest M, et al. Optimal Intensity and Duration of Walking Rehabilitation in Patients With Chronic Stroke: A Randomized Clinical Trial. JAMA Neurol. 2023;80:342–351.
Cousineau D, Chartier S. Outliers detection and treatment: A review. Int. J. Psychol. Res. 2010;3.
Cooper S, Smayda K, et al. InTandem™, a neurorehabilitation system to improve walking in the chronic stroke population: A validation study assessing safety and effectiveness of use. JMIR Rehabil Assist Technol (forthcoming). doi:10.2196/50438 http://dx.doi.org/10.2196/50438

Table 1. Baseline Demographic and Clinical Characteristics of Participants

Descriptive Variable	Cohort A InTandem (n=40)	Cohort B Active Control (n=32)	Overall (n=72)	Sig. (p)
Age
Mean (SD)	61.3 (6.5)	63.6 (7.5)	62.3 (7.0)	.168
Gender				.200
Male	26 (65.0%)	16 (50.0%)	42 (58.3%)
Female	14 (35.0%)	16 (50.0%)	30 (41.7%)
Ethnicity				.797
Hispanic or Latino	5 (12.5%)	2 (6.3%)	7 (9.7%)
Not Hispanic or Latino	31 (77.5%)	27 (84.4%)	58 (80.6%)
Unknown	2 (5.0%)	2 (6.3%)	4 (5.6%)
Not Reported	2 (5.0%)	1 (3.1%)	3 (4.2%)
Race
American Indian or Alaska Native	0 (0.0%)	0 (0.0%)	0 (0.0%)
Asian	0 (0.0%)	1 (3.1%)	1 (1.4%)
Black or African American	16 (40.0%)	15 (46.9%)	31 (43.1%)
Native Hawaiian or Other Pacific Islander	0 (0.0%)	0 (0.0%)	0 (0.0%)
White	17 (42.5%)	14 (43.8%)	31 (43.1%)
Identified as more than 1 race	2 (5.0%)	0 (0.0%)	2 (2.8%)
Unknown	2 (5.0%)	0 (0.0%)	2 (2.8%)
Not Reported	3 (7.5%)	2 (6.3%)	5 (6.9%)
Education Level				.894
Some High School	3 (7.5%)	1 (3.1%)	4 (5.6%)
High School Graduate	7 (17.5%)	6 (18.8%)	13 (18.1%)
Some College	14 (35.0%)	11 (34.4%)	25 (34.7%)
College Graduate	12 (30.0%)	9 (28.1%)	21 (29.2%)
Graduate Degree	4 (10.0%)	5 (15.6%)	9 (12.5%)
Time Since Stroke (years)				.684
Mean (SD)	8.4 (6.9)	7.7 (7.6)	8.1 (7.1)

Table 2. Relatedness and severity of reported adverse events

Adverse Event Relatedness and Severity	InTandem	Control
Not related
Mild
covid (infected)	2
covid vaccine	1
fall	1	4
headache		1
hypertension	1
pain	2
tonsillitis	1
Moderate
hospitalization		1*
Severe
weakness	1*
Life-threatening
seizures		1*
Unlikely Related
Mild
fall	1
Possibly Related
Mild
hypertension	1
Severe
chest pain/diaphoresis/tachycardia	1*
Probably Related
Mild
pain		5
Grand Total	12	12

* Denotes a Serious Adverse Event (SAE)

Yes there is potential Competing Interest. LNA is a paid advisor to MedRhythms Inc. BH is cofounder and CEO of MedRhythms Inc with equity interest. ST and DP are employees of MedRhythms Inc with equity interest. This study was completed under a master management plan instituted by the Boston University Financial Conflicts of Interest Committee.

SupplementaryMaterialAppendixAandB.docx
Supplementary Material_Appendix A and B

Download PDF

Journal Publication

published 08 Feb, 2024

Read the published version in Nature Communications →

Version 1

posted

You are reading this latest preprint version

Efficacy and safety of using auditory-motor entrainment to improve walking after stroke: A multi-site randomized controlled trial of the InTandem^TMneurorehabilitation system

Status:

Journal Publication

Version 1

Abstract

Figures

INTRODUCTION

METHODS

Inclusion and exclusion criteria

Data Collection Overview

Primary Endpoint Data Collection

Randomization

Trial Design Overview

InTandem System and Intervention

Active Control

COVID–19 Considerations

Statistical Analyses

RESULTS

Participants

Session completion and InTandem system reliability

Primary Endpoint Analyses

Exploratory Time-Course of Change Analysis

Safety: Frequency of adverse events

DISCUSSION

Faster walking speeds and rate of recovery

Faster walking, but not at the expense of safety

Limitations

Conclusions

Declarations

References

Tables

Additional Declarations

Supplementary Files

Status:

Journal Publication

Version 1