Skip to main content
NCBI home page
Sage Choice logoLink to Sage Choice
. 2025 Dec 23;44(2):55–67. doi: 10.1177/09226028251392269

Ten Recommendations for Improving Research on Stroke Motor Rehabilitation: A Unified Perspective

Chandramouli Krishnan 1,2,✉, Thomas E Augenstein 3, Rajiv Ranganathan 4
Editors: Chandramouli Krishnan, James Sulzer, Carolynn Patten, Rajiv Ranganathan
PMCID: PMC12967000  NIHMSID: NIHMS2147833  PMID: 41432691

Abstract

Stroke motor rehabilitation research is critical for improving functional recovery and quality of life. However, the success of therapeutic interventions is highly dependent on the theoretical basis and methodological rigor of the studies that evaluate them. This paper outlines ten essential recommendations/guidelines for stroke motor rehabilitation research aimed at enhancing the scientific quality and reliability of findings. Our recommendations span issues related to mechanistic understanding, methodological rigor, and transparency, while providing specific suggestions for implementation in the context of stroke motor rehabilitation research. To improve our mechanistic understanding of how and why an intervention works, these guidelines emphasize the importance of having a clear scientific rationale for the active ingredient in the proposed therapy (e.g., physical or occupational therapy), ensuring manipulation checks for key therapeutic components, and assessing outcomes at multiple levels using the ICF framework. To improve methodological rigor, the paper also highlights the necessity of adequately dose-matched control groups, minimizing baseline imbalances and biased treatment effects through randomization and blinding, and employing larger sample sizes to minimize the risk of false-positive trials. Finally, to enhance transparency and reproducibility, we advocate for pre-registering outcomes and protocols, performing robustness checks, presenting data in multiple formats, and the publication of open data. We anticipate that following these recommendations will pave the way for more reliable, impactful results that advance the development of effective therapeutic interventions.

Keywords: hemiparesis, physical therapy, occupational therapy, scientific rigor, clinical trial, drug, device, machine learning, artificial intelligence

Ten Recommendations for Stroke Motor Rehabilitation Research

Stroke survivors continue to experience significant residual motor and functional deficits even several months after stroke, highlighting a critical need to identify effective therapeutic approaches for motor rehabilitation (Winstein et al., 2016). However, evaluating the effectiveness of stroke rehabilitation therapies is contingent on research rigor and transparency. Despite many promising pilot and early-stage clinical interventions, the number of successful Phase 3 clinical trials (i.e., trials that show significantly better outcomes for the tested interventions when compared with the control interventions) remains relatively low (Stinear et al., 2020). This points to the need for developing better guidelines for conducting research in this area.

In this opinion paper, we provide a set of key recommendations/guidelines for stroke motor rehabilitation research, focusing on the scientific rigor, methodological integrity, and transparency necessary to advance knowledge and discovery. For each guideline, we highlight its importance and suggest examples of how it could be implemented in stroke motor rehabilitation research. Although many of these issues have been discussed individually in previous work [e.g., (Borschmann et al., 2018; Ranganathan et al., 2022a; Ranganathan et al., 2022b; Stinear et al., 2020; Tsay & Winstein, 2021; Van Stan et al., 2019; Van Stan et al., 2021; Van Stan et al., 2023; Winstein et al., 2016; Winstein & Kay, 2015)], we believe that synthesizing them into a set of recommendations will assist both researchers and clinicians in designing and conducting well-designed experiments that will advance stroke motor rehabilitation.

Guideline 1: Specify a Clear Scientific Rationale for the Active Ingredient(s) of Therapy

The “active ingredient” refers to the therapeutic component of the intervention that is intended to produce a clinical effect (Darekar, 2023; Hart & Ehde, 2015; Johnson et al., 2021; Ranganathan et al., 2022a). Interventions in stroke motor rehabilitation tend to be very diverse (behavioral, pharmacological, and neurostimulation) with different theoretical bases and multiple active ingredients (Whyte, 2009). Therefore, it is essential to provide an explicit and well-supported scientific rationale for “how” the chosen active ingredient is expected to work (Ranganathan et al., 2022a; Tsay & Winstein, 2021). Examples of active ingredients that are typically used in stroke motor rehabilitation are provided in Figure 1.

Figure 1.

Figure 1.

Open in a new tab

A schematic of common active ingredients used in therapeutic interventions.

Importance of Active Ingredients

Without the specification of active ingredients, the mechanism of action in the intervention is unclear. This information is critical from the scientific standpoint of understanding and generalizing results beyond the specific study. The lack of clear active ingredients can often lead to ‘proxy’ (and potentially misleading) labeling of studies based on other factors such as technology (e.g., a “robotics” trial, or a “VR” trial), which distracts from what the technology is designed to do. In addition to providing a clear rationale, the specification of the active ingredient is critical for assessing alignment with the manipulation checks (as described in guideline #2) and the measurement of outcomes (guideline #3).

Implementing Active Ingredients in Stroke Motor Rehabilitation Trials

Specifying active ingredients for an intervention requires triangulation of evidence from many sources. For example, if a study is investigating the use of transcranial direct current stimulation (tDCS) to enhance motor recovery post-stroke, the scientific rationale could include evidence from neuroscience, such as the effects of tDCS on cortical excitability, neural plasticity, and motor function recovery. The rationale should demonstrate how this therapy might be effective for stroke survivors, based on prior studies, animal models, or theoretical frameworks (Ranganathan et al., 2022a). Given the prevalence of mixed results in the literature, while it is easy to find articles that support one's theoretical stance, it is important to fully evaluate the rigor of prior research by carefully assessing both positive and negative findings.

Active ingredients are critical even in cases where there is not a ‘single’ mechanism that is key to recovery. For example, a multimodal intervention combining neurostimulation with motor training should include a rationale for how the active ingredient in each intervention complements each other so that their combination can synergistically promote rehabilitation. This approach not only prevents ad-hoc combinations from being tested without a clear scientific rationale (Ranganathan et al., 2022a) but can also provide a basis for understanding the complex multifactorial nature of rehabilitation.

Guideline 2: Include Manipulation Checks for the Key Active Ingredient to Ensure That the Therapy Indeed Affected the Key Ingredient

Manipulation checks are assessments that verify whether the intervention successfully altered the proposed target (Bedwell et al., 2023; Ejelöv & Luke, 2020; Hauser et al., 2018; Hoewe, 2017). Once the active ingredient has been identified, it is essential to include manipulation checks to ensure that the therapy is targeting the intended mechanism of action. A successful manipulation check is a critical part of the evidence for the mechanism and avoids the creation of ‘phantom’ active ingredients that are proposed as explanations without being directly tested.

Importance of Manipulation Checks

Without manipulation checks, there is no evidence that the intervention engaged the intended therapeutic mechanism, which in turn, greatly reduces the interpretability of a study. For instance, if a study investigates the effects of a neurostimulation protocol intended to increase corticospinal excitability (the active ingredient), manipulation checks should confirm that corticospinal excitability indeed has increased. When the manipulation check does not agree with the eventual outcomes (i.e., a failed manipulation check in a study that shows successful outcomes or a successful manipulation check in a study with failed outcomes), this can be critical information for discovering alternative mechanisms of action (Ranganathan et al., 2022a).

Conducting Manipulation Checks

Manipulation checks for stroke motor rehabilitation trials vary depending on the type of intervention and the underlying active ingredient. For behavioral therapies like strength or motor control training, performance measures, such as force production or variability/error, can ensure that the therapy is indeed improving strength and motor control as expected (Patel et al., 2021; Sharp & Brouwer, 1997; Weiss et al., 2000). For neurostimulation, this can be done using measures such as transcranial magnetic stimulation (TMS) to evaluate changes in motor cortical excitability before and after the intervention (Abbruzzese & Trompetto, 2002) For pharmacological therapies, blood biomarkers or imaging modalities (e.g., MRI or CT scans) can provide evidence that the drug has engaged its intended biological target (Bilgic et al., 2020; Cramer, 2015; Tardy et al., 2006).

Guideline 3: Assess Outcomes at Multiple Levels (Biomechanical, Neurophysiological, Clinical) of the ICF Framework

As stroke recovery often involves improvements across multiple facets of the individual's life (Tsay & Winstein, 2021), it is important to assess outcomes at multiple levels to capture the full range of rehabilitation effects. The International Classification of Functioning, Disability, and Health (ICF) model is an excellent framework for structuring these outcomes (Kwakkel et al., 2017; Schepers et al., 2007). The ICF includes three domains: impairments (e.g., weakness, spasticity), activity limitations (e.g., difficulty walking), and participation restrictions (e.g., social isolation). Researchers should consider using outcomes that provide insight into how the proposed therapy affects each of the ICF domains.

Importance of Quantifying Outcomes at Multiple Levels

Without assessments of outcomes at multiple levels, it is difficult to capture the true effect of a rehabilitation intervention and the mechanisms of recovery. For example, because of the redundancy in the motor system, improvements in reaching might be caused by trunk compensation instead of actual recovery of the arm (Cirstea & Levin, 2000; Jayasinghe et al., 2021). Similarly, the time course of improvements at different levels is different; biomechanical and neurophysiological outcomes often (but not always) show changes that precede subsequent improvements at the activity and participation levels. For instance, enhanced corticospinal excitability after a rehabilitation intervention may be indicative of neural reorganization and may predict future gains in motor function (Tarkka et al., 2008). Therefore, multi-level assessments are key to identifying these cascading mechanisms driving recovery.

Integrating Multi-Level Outcome Assessments

Experiments should consider not just clinical measures, such as improvements in function or activities of daily living (ADLs), but also biomechanical and neurophysiological changes that might occur as a result of therapy (Krishnan et al., 2012; Kwakkel et al., 2017). For example, in a study investigating the effectiveness of a gait rehabilitation protocol, outcome measures might include changes in muscle strength and gait mechanics (biomechanical), motor cortical excitability (neurophysiological), and functional performance measures such as gait speed and endurance (clinical) (Krishnan et al., 2012). Further, such measurement of outcome measures at multiple levels should consider the use of standardized outcome measures that have been previously recommended by stroke motor rehabilitation experts (Kwakkel et al., 2017). Such a multi-dimensional approach to assessment will provide a more comprehensive understanding of how interventions influence recovery.

Guideline 4: Choosing Appropriate ‘Dose-Matched’ Control Groups

In any experimental design, the control group should be carefully matched to the experimental group to ensure that any differences observed can be attributed to the intervention rather than other confounding variables. For stroke motor rehabilitation trials, one important factor in this aspect is “dose”—i.e., ensuring that the control group receives an intervention that is equivalent to the experimental group but without the key active ingredient. Unlike other types of trials (e.g., pharmaceutical drug trials) where such a placebo-controlled group is relatively easy to implement, this issue requires greater attention in stroke motor rehabilitation trials from scientific, pragmatic, and ethical perspectives.

Importance of a Dose-Matched Control Group

Without a dose-matched control group, the efficacy of a treatment and the underlying mechanisms of action can be unclear. Because dose is a multidimensional construct (Hayward et al., 2021), matching dose across all of these dimensions can be challenging. As a result, a commonly used alternative that is justifiable from a pragmatic and ethical standpoint is the inclusion of a “usual care” control group. However, this type of control group presents notable limitations. Usual care in stroke motor rehabilitation is inherently heterogeneous in its content and lacks a standardized structure (Arienti et al., 2022; Negrini et al., 2020; Rodgers et al., 2020). Such heterogeneity in implementation and reporting of the control group affects both internal and external validity (Lohse et al., 2018). Furthermore, since it is likely that virtually any structured therapy produces better outcomes than no therapy, comparing an experimental intervention to usual care (especially if usual care involves minimal therapy) can result in misleading conclusions about efficacy.

Designing Control Groups to Isolate Active Ingredients

Although pragmatic constraints are important in some contexts since they enhance the likelihood of future implementation, it is equally important to consider ambitious (e.g., very high dose), novel interventions that challenge existing paradigms and push the field forward (Krakauer & Carmichael, 2022; Lin et al., 2025; Ward, 2018). Therefore, rather than using a ‘usual care’ control group as a default, the control group should be designed based on the specific research question and the intended mechanism. For example, if the experiment involves testing the effects of split-belt treadmill training (where participants walk with one leg moving much slower than the other, usually at half or one-third of the speed), the control group could be a dose-matched treadmill walking group or a dose-matched split-belt treadmill training at a coupling speed that is known to have very minimal or no effect (e.g., one leg moving at 90% of the speed of the other leg). This allows researchers to isolate the effects of the active ingredient (i.e., the asymmetry in gait) and rule out other factors that could contribute to the observed outcomes (such as the overall dose in terms of walking duration).

Guideline 5: Perform Appropriate Randomization to Minimize Baseline Imbalances

Randomization is the process of assigning participants to different treatment groups such that each participant has an equal chance of receiving any treatment. Randomization is an essential component of a rigorous clinical trial design and helps ensure that participants are assigned to experimental and control groups without bias (Bridgman et al., 2003; Lim & In, 2019). Selecting an appropriate randomization method is essential in rehabilitation research to ensure valid comparisons and minimize bias, especially when baseline characteristics may influence outcomes.

Importance of Randomization and Minimizing Baseline Imbalances

Assigning participants randomly to a group increases the likelihood that both known and unknown characteristics (e.g., age, severity of stroke, or comorbidities) are evenly distributed across groups, minimizing imbalances between treatment groups at the start of the trial (i.e., baseline differences). This is critical because significant differences in participant characteristics at the start of a trial can distort treatment effects and lead to incorrect conclusions. While statistical adjustments (e.g., ANCOVA) can partially correct for these imbalances, they cannot fully compensate for poor randomization and could confound interpretability. Therefore, proper randomization and careful monitoring of baseline imbalances are key to ensuring the internal validity and interpretability of clinical trial results.

Randomization Strategies to Minimize Bias

Selecting an appropriate randomization method is essential in rehabilitation research to ensure valid comparisons and minimize bias, especially when baseline characteristics may influence outcomes. In large trials, simple randomization is appropriate, even if it results in uneven group sizes. However, in trials involving acute or sub-acute stroke survivors where the baseline characteristics of the participants could affect their ability to participate in the treatment (e.g., ability to walk in a trial that is testing a locomotor intervention), stratified block randomization might be more appropriate to allocate participants (e.g., as in the LEAPS trial) (Duncan et al., 2011). For studies with smaller sample sizes, stratified block or adaptive randomization (Crawford et al., 2024; Hayward et al., 2024) is preferable to balance key confounding variables while maintaining group comparability. Regardless of the randomization method, ensuring baseline equivalence between groups is crucial to attributing outcomes to the intervention rather than preexisting differences, and any imbalances should be thoroughly evaluated.

Guideline 6: Ensure Adequate Blinding at All Levels

Blinding (also known as masking) is a methodological strategy used to prevent participants, researchers or clinical providers, and/or outcome assessors or data analysts from knowing the group assignments of participants. Similar to randomization, blinding is an essential component of a clinical trial (Bridgman et al., 2003; Hrobjartsson et al., 2014; Jadad et al., 1996) and minimizes the risk of observer or participant bias influencing the results.

Importance of Blinding

Without blinding, it is difficult to distinguish whether perceived improvements are due to placebo effects/observer bias or true intervention efficacy. While randomization minimizes bias at the start of the trial, blinding minimizes biased treatment effects due to differential treatment of the groups or assessment of outcomes later in the trial (Karanicolas et al., 2010). Blinding, particularly of assessors and participants, is critical in reducing both conscious and unconscious influences on outcomes. The use of standardized protocols, automated measures, and separate personnel for treatment and assessment can help enforce this separation.

Strategies for Blinding

Blinding in stroke motor rehabilitation trials can be challenging, especially when the intervention involves the use of technologies such as functional electrical stimulation (FES) or robotics, as participants and therapists are aware of the treatment they are receiving. However, even when treatments differ visibly or physically, researchers can still reduce expectancy effects. For example, in sham-controlled studies using technologies like TMS or FES, placebo conditions should be designed to mimic the sensory (e.g., sham TMS coils that mimic the sound without inducing magnetic fields (Ruohonen et al., 2000) experience of active treatment. In behavioral interventions, expectancy effects can be minimized by providing neutral framing by avoiding suggestive language and using standardized scripts, and limiting discussion of group allocation. When participant blinding is challenging, at least the assessors should be blinded to treatment allocation, ensuring that outcome assessments are unbiased. Including these layers of blinding greatly strengthens the methodological rigor and credibility of stroke rehabilitation research.

Guideline 7: Use Larger Sample Sizes to Reduce the Risk of False-Positive Results

Sample size is the total number of participants included or analyzed in a study. In stroke rehabilitation, where the variability between participants is high (Louw, 2002) and clinical improvements tend to be modest, small sample sizes can significantly undermine the reliability, validity, and generalizability of the results.

Importance of Large Sample Sizes

Although the idea that small sample sizes result in low power (making it more likely to miss true effects) is fairly well known, a common misconception is that a statistically significant result in a small sample implies a strong or robust treatment effect (also referred to as the “what does not kill my statistical significance makes it stronger” fallacy (Gelman, 2017). It is critical to note that small sample sizes also decrease the chances that an observed effect is a true effect, and inflate the effect sizes when there is indeed a true effect (Button et al., 2013; Forstmeier et al., 2017; Schmidt et al., 2014). This is one reason that it is preferable to scale down the estimated effect size in power calculations based on prior work (Norris et al., 2024). In addition to these issues, larger sample sizes that include a more representative population also ensure that results are more generalizable to the broader population of stroke survivors.

Balancing Feasibility and Validity When Choosing Sample Sizes

Given the heterogeneity of stroke and the strict inclusion criteria for many interventions, large sample sizes in stroke motor rehabilitation are not always feasible, particularly in early-phase trials, studies involving rare subpopulations, or resource-limited settings. Researchers should consider collaborating and pooling data from multiple labs to improve sample size, especially when there are financial constraints and recruitment challenges. Importantly, small studies should be framed as preliminary or hypothesis-generating, and their results interpreted cautiously until larger, confirmatory trials are conducted. In addition, researchers can take additional steps (see recommendations 8–10) to strengthen rigor and mitigate the risk of false interpretations.

It is also worth noting that what is considered a ‘large’ or ‘small’ sample size is not an absolute number but is inherently tied to the research question and the effect size. For example, there is a growing emphasis on conducting trials during the acute or sub-acute phases of stroke, when neuroplasticity is at its peak and the brain is most responsive to rehabilitation (Hordacre et al., 2021). However, these early phases are also marked by highly variable recovery trajectories. Conducting small-sample trials in this phase heightens the risk of misattributing natural recovery to treatment effects, especially if randomization fails to balance key prognostic factors. Therefore, small studies in this context should be approached with extreme caution, unless designed explicitly for hypothesis generation and conducted with rigorous methods such as stratified randomization based on some valid predictive models (e.g., presence or absence of motor evoked potentials induced by transcranial magnetic stimulation) (Smith et al., 2024; Stinear et al., 2017).

Guideline 8: Pre-Register Outcomes and Protocols

Pre-registration involves publicly documenting the study's hypothesis, outcomes, and methodology before data collection begins. By committing to a predefined set of outcomes, research questions, and data analytic plan, researchers are held accountable to the protocol they have established, thereby ensuring transparency and reproducibility (Munro & Prendergast, 2019; P Simmons et al., 2021).

Importance of Pre-Registration

Without pre-registration, there is no constraint on ‘researcher degrees of freedom’ (Simmons et al., 2011). Flexibility in terms of exploring multiple outcomes, multiple analysis procedures and selective reporting can blur the distinction between ‘confirmatory’ and ‘exploratory’ research, leading to inflated false-positive rates and undermining scientific rigor (Cook et al., 2025; Munro & Prendergast, 2019; Simmons et al., 2011). Pre-registering studies helps constrain such degrees of freedom through greater transparency, as it allows others to see what decisions were planned in advance and to assess whether any changes were made during the course of the study (e.g., by looking at the record history in clinicaltrials.gov). This transparency contributes to the credibility and reproducibility of the research.

Implementing Transparency in Stroke Motor Rehabilitation Research

Pre-registration is critical in stroke motor rehabilitation, where clinical trials often involve complex interventions with multiple outcome measures. Pre-registering experimental protocols and analysis plans, either as registered reports in academic journals or on platforms like ClinicalTrials.gov, is crucial to minimize questionable research practices like p-hacking, outcome switching, and selective reporting (Brodeur et al., 2024; Cook et al., 2025; Head et al., 2015; P Simmons et al., 2021; Simmons et al., 2011). It is to be noted that while the International Committee of Medical Journal Editors (ICMJE) mandates prospective registration of clinical trials in publicly accessible registries, there are several well-known limitations of both data and practical usage of platforms like clinicaltrials.gov (Miron et al., 2020; Tse et al., 2018; Turner et al., 2021). The use of registered reports (which go beyond pre-registration) can potentially offer a more rigorous process, requiring detailed methodological and statistical analysis plans to be peer-reviewed and approved before data collection begins, enhancing transparency and reducing the risk of selective reporting. Furthermore, registered reports help combat publication bias by encouraging the publication of null results, which is critical for unbiased systematic reviews and meta-analyses (Song et al., 2013).

Guideline 9: Perform Robustness Checks and Present Data with Greater Transparency

Robustness checks refer to transparently running multiple versions of data analysis (i.e., multiverse analysis) to confirm that the result is not sensitive to minor changes in the assumptions or methodology (e.g., normalization methods, outlier exclusion) (Dragicevic et al., 2019; Ranganathan et al., 2022b; Steegen et al., 2016). Transparency refers to presenting data in multiple formats to allow for a complete understanding of the results. Data visualization tools such as raincloud plots, which combine raw data points, boxplots, and probability density estimates to illustrate both the distribution and individual variability in the data, are preferable to simpler summary statistics or bar graphs that can obscure distribution and variability in the underlying data (Allen et al., 2019). Similar to pre-registration, both robustness checks and visualization address the issue of increasing transparency and credibility by ensuring that the conclusions drawn are not an artifact of a specific (potentially arbitrary) choice made by the researcher.

Importance of Robustness Checks and Data Transparency

Robustness checks help confirm that the findings are reliable and not the result of data processing artifacts or statistical issues. Reporting robustness checks transparently strengthens the credibility of the findings and builds confidence in the intervention's effectiveness. Presenting results in multiple formats (e.g., raw data tables, summary statistics, and visualizations) enhances the interpretability, transparency, and reproducibility of findings (Klein et al., 2018; Weissgerber et al., 2019). Data visualization that shows individual-level data (e.g., boxplots with individual data points, raincloud plots, individual trajectories) allows readers to observe trends, variability, and outliers that may not be evident in standard statistical summaries (such as bar plots with mean and standard errors).

Ensuring Robustness of Analysis and Results

Intervention effects in stroke survivors are often highly variable, and group averages can obscure clinically meaningful individual responses to therapy. Robustness checks may include testing different data normalization techniques and/or performing sensitivity analyses by excluding outliers or specific subgroups (Dragicevic et al., 2019; Krishnan et al., 2024; Steegen et al., 2016). Presenting changes in outcomes over time at the individual level (e.g., spaghetti plots) can help contextualize group findings and illustrate heterogeneity in treatment effects (see for example (Waddell et al., 2019)).

Robustness checks are also critical for machine learning (ML) models that are becoming increasingly used in stroke motor rehabilitation research to predict individual differences in recovery trajectories and therapy response, offering the potential to move beyond group averages toward more personalized care (Campagnini et al., 2025; Finocchi et al., 2024; Harari et al., 2020; Kim et al., 2022; Kuch et al., 2024; Nicora et al., 2025; Quattrocelli et al., 2024; Winner et al., 2024). In the context of applying ML to this particular domain, robustness concerns arise from the limitations of the data itself: (i) sample sizes are limited with multiple heterogenous data streams (e.g., neuroimaging, wearable sensors, clinical scores, patient-reported outcomes), which can increase the risk of overfitting, particularly when using complex models, (ii) clinical datasets are frequently noisy and incomplete (e.g., missing assessments, variability in clinician input, or patient dropout), which can lead to bias, and (iii) the lack of independent data sets for estimating model generality, i.e., many models are trained and tested on the same dataset (usually from a single location) without evaluation on independent cohorts from different labs or clinics (Garcia-Rudolph et al., 2021; Zu et al., 2023). Therefore, robustness checks, such as assessing model stability across subgroups, imputing or excluding missing data, and transparently reporting prediction intervals, are essential to ensure that the insights derived from ML models are both scientifically valid and clinically meaningful.

Guideline 10: Provide Open Data

Open data refers to the practice of publishing data on platforms where other researchers can access the data. Rehabilitation scientists should consider publishing raw data and using open-science platforms (e.g., https://osf.io/) to share their findings.

Importance of Open Data

Sharing raw or minimally processed data, whether in appendices or online repositories, promotes open science practices and enables independent reanalysis or meta-analytic integration by other researchers (Mayernik, 2017; Murray-Rust, 2008). Further, open data enables researchers not only to verify findings using the same data sets but also to improve efficiency by reducing the need to collect new data repeatedly. Open data also allows innovative methodologies such as individual participant data (IPD) meta-analysis, which is more robust than traditional meta-analysis that uses summary-level data (Riley et al., 2010).

Implementing Open Data in Stroke Motor Rehabilitation Research

Implementing open data in stroke motor rehabilitation research involves creating accessible, standardized datasets that can be shared across institutions and researchers. Open data not only allows others to replicate findings but also fosters collaboration and accelerates scientific progress. For instance, ATLAS v2.0 is a comprehensive archive of T1-weighted stroke MRI images (n = 1271) and manually segmented lesion masks that include training, test, and generalizability datasets (Liew et al., 2022). This resource has the potential to facilitate the development of advanced lesion segmentation algorithms, significantly advancing stroke motor rehabilitation research. Similarly, a large harmonized dataset of accelerometry data collected over the past 10 + years has also been shared recently (Lang, 2024a, 2024b). This dataset includes 24-h real-world accelerometry data from the upper and lower limbs of both stroke survivors and neurologically intact adults, as well as demographic and clinical information about the participants. This publicly available dataset could enable the development of predictive models for real-life limb usage in stroke survivors. Thus, by making data and methodology publicly available, researchers not only enhance the credibility of their work but also contribute to the collective knowledge base and the advancement of rehabilitation science.

There are also ethical issues with open data—so it is important to recognize that this requires adequate planning at the time of the institutional review board approval, so that data can be obtained and shared ethically with informed consent from participants. When making the data publicly available, researchers should also ensure patient privacy through de-identification protocols and use interoperable formats (e.g., CSV, JSON, or XML) for easy integration across different platforms. Researchers should publish datasets in recognized repositories with clear metadata, promoting transparency, collaboration, and reproducibility.

Summary

In conclusion, the ten recommendations/guidelines provided in this manuscript (Table 1) are crucial for ensuring the rigor, transparency, and reliability of stroke motor rehabilitation research. While these recommendations are discussed in the context of stroke motor rehabilitation research, many of these concepts and guidelines could apply broadly to rehabilitation research in general. It is important to acknowledge that performing a stroke motor rehabilitation study is not easy and requires immense effort in planning, coordination, recruitment, and implementation. By following these recommendations that emphasize careful design, robust methodologies, and transparent reporting, we hope that such efforts contribute more effectively to the development of transformative therapies for stroke survivors.

Table 1.

Ten Guidelines for Stroke Motor Rehabilitation Research.

Purpose # Guideline Key Elements Outcome
To Improve Mechanistic Understanding of Why and How an Intervention works 1 Have a Clear Scientific Rationale for the Active Ingredient Define the mechanism of action; avoid ad-hoc multimodal approaches Establishes the theoretical basis and expected outcomes of therapy
2 Include Manipulation Checks Verifies therapy affects the intended target; types of checks (behavioral, physiological) Confirms intervention targets the hypothesized mechanism
3 Assess Outcomes at Multiple Levels Using ICF Models Clinical, biomechanical, neurophysiological; use of International Classification of Functioning (ICF) framework Captures the holistic effects of an intervention
To Improve Methodological Rigor 4 Include Dose-Matched, Well-Controlled Comparison Groups Control group matched in time/intensity except for active ingredient Isolates the true effect of the intervention
5 Randomize to Minimize Baseline Imbalances Perform appropriate randomization to ensure similar baseline characteristics; avoid Lord's Paradox Improves internal validity and prevents confounding
6 Ensure Adequate Blinding Blinding at multiple levels (e.g., participants, interventionists, assessors); address challenges specific to rehabilitation protocols and settings Reduces observer and participant bias
7 Use Larger Sample Sizes Power analysis; pool data from multiple labs; reduce the risk of false positives Increases statistical power and generalizability
To Enhance Transparency and Reproducibility 8 Pre-register Outcomes and Protocols Registered reports or clinical trial registration; public documentation of hypotheses, outcomes, and methods Enhances transparency and reduces bias
9 Check for Robustness and Enhance Data Transparency Conduct robustness checks/multiverse analyses; present data in multiple formats, statistical tables, robust visualizations (e.g., raincloud plots) Strengthens interpretability and replicability
10 Open Data Open-access data (e.g., OSF); publication of null results Promotes reproducibility and reduces publication bias
Open in a new tab

Acknowledgements

This work was supported in part by the Eunice Kennedy Shriver National Institute of Child Health and Human Development, National Institutes of Health (Grant # R41-HD111289 and R01-HD111567).

Footnotes

ORCID iDs: Chandramouli Krishnan https://orcid.org/0000-0002-7278-7389

Thomas E Augenstein https://orcid.org/0000-0002-2781-3066

Rajiv Ranganathan https://orcid.org/0000-0002-6924-253X

Funding: The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by the Clinical Center, (grant number R01-HD111567, R41-HD111289 ).

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

References

  1. Abbruzzese G., Trompetto C. (2002). Clinical and research methods for evaluating cortical excitability. Journal of Clinical Neurophysiology, 19(4), 307–321. 10.1097/00004691-200208000-00005 [DOI] [PubMed] [Google Scholar]
  2. Allen M., Poggiali D., Whitaker K., Marshall T. R., van Langen J., Kievit R. A. (2019). Raincloud plots: a multi-platform tool for robust data visualization. Wellcome Open Res, 4, 63. 10.12688/wellcomeopenres.15191 [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Arienti C., Buraschi R., Pollet J., Lazzarini S. G., Cordani C., Negrini S., Gobbo M. (2022). A systematic review opens the black box of “usual care” in stroke rehabilitation control groups and finds a black hole. Eur J Phys Rehabil Med, 58(4), 520–529. 10.23736/S1973-9087.22.07413-5 [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Bedwell G. J., Chikezie P. C., Siboza F. T., Mqadi L., Rice A. S. C., Kamerman P. R., Parker R., Madden V. J. (2023). A systematic review and meta-analysis of non-pharmacological methods to manipulate experimentally induced secondary hypersensitivity. The Journal of Pain, 24(10), 1759–1797. 10.1016/j.jpain.2023.06.013 [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Bilgic A. B., Gocmen R., Arsava E. M., Topcuoglu M. A. (2020). The effect of clot volume and permeability on response to intravenous tissue plasminogen activator in acute ischemic stroke. Journal of Stroke and Cerebrovascular Diseases, 29(2), 104541. 10.1016/j.jstrokecerebrovasdis.2019.104541 [DOI] [PubMed] [Google Scholar]
  6. Borschmann K., Hayward K. S., Raffelt A., Churilov L., Kramer S., Bernhardt J. (2018). Rationale for Intervention and Dose Is Lacking in Stroke Recovery Trials: A Systematic Review. Stroke Res Treat, 13, 8087372. 10.1155/2018/8087372 [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Bridgman S., Engebretsen L., Dainty K., Kirkley A., Maffulli N., Committee I. S. (2003). Practical aspects of randomization and blinding in randomized clinical trials. Arthroscopy: The Journal of Arthroscopic & Related Surgery, 19(9), 1000–1006. 10.1016/j.arthro.2003.09.023 [DOI] [PubMed] [Google Scholar]
  8. Brodeur A., Cook N. M., Hartley J. S., Heyes A. (2024). Do preregistration and preanalysis plans reduce p-hacking and publication bias? Evidence from 15,992 test statistics and suggestions for improvement. Journal of Political Economy Microeconomics, 2(3), 527–561. 10.1086/730455 [DOI] [Google Scholar]
  9. Button K. S., Ioannidis J. P., Mokrysz C., Nosek B. A., Flint J., Robinson E. S., Munafo M. R. (2013). Power failure: Why small sample size undermines the reliability of neuroscience. Nature Reviews Neuroscience, 14(5), 365–376. 10.1038/nrn3475 [DOI] [PubMed] [Google Scholar]
  10. Campagnini S., Sodero A., Baccini M., Hakiki B., Grippo A., Macchi C., Mannini A., Cecchi F. (2025). Prediction of the functional outcome of intensive inpatient rehabilitation after stroke using machine learning methods. Scientific Reports, 15(1), 16083. 10.1038/s41598-025-00781-1 [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Cirstea M. C., Levin M. F. (2000). Compensatory strategies for reaching in stroke. Brain, 123(Pt 5), 940–953. 10.1093/brain/123.5.940 [DOI] [PubMed] [Google Scholar]
  12. Cook B. G., Wong V. C., Fleming J. I., Solari E. J. (2025). Preregistration of randomized controlled trials. Research on Social Work Practice, 35(3), 277–286. 10.1177/10497315221121117 [DOI] [Google Scholar]
  13. Cramer S. C. (2015). Drugs to enhance motor recovery after stroke. Stroke, 46(10), 2998–3005. 10.1161/STROKEAHA.115.007433 [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Crawford A. M., Lorenzi E. C., Saville B. R., Lewis R. J., Anderson C. S. (2024). Adaptive clinical trials in stroke. Stroke, 55(11), 2731–2741. 10.1161/STROKEAHA.124.046125 [DOI] [PubMed] [Google Scholar]
  15. Darekar A. (2023). Virtual reality for motor and cognitive rehabilitation. Curr Top Behav Neurosci, 18(65), 337–369. 10.1007/7854_2023_418 [DOI] [PubMed] [Google Scholar]
  16. Dragicevic P., Jansen Y., Sarma A., Kay M., Chevalier F. (2019). Increasing the transparency of research papers with explorable multiverse analyses, proceedings of the 2019 chi conference on human factors in computing systems, pp. 1–15.
  17. Duncan P. W., Sullivan K. J., Behrman A. L., Azen S. P., Wu S. S., Nadeau S. E., Dobkin B. H., Rose D. K., Tilson J. K., Cen S., Hayden S. K., Team L. I. (2011). Body-weight-supported treadmill rehabilitation after stroke. New England Journal of Medicine, 364(21), 2026–2036. 10.1056/NEJMoa1010790 [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Ejelöv E., Luke T. J. (2020). “Rarely safe to assume”: Evaluating the use and interpretation of manipulation checks in experimental social psychology. Journal of Experimental Social Psychology, 18(11), 103937. 10.1016/j.jesp.2019.103937 [DOI] [Google Scholar]
  19. Finocchi A., Campagnini S., Mannini A., Doronzio S., Baccini M., Hakiki B., Bardi D., Grippo A., Macchi C., Navarro Solano J., Baccini M., Cecchi F. (2024). Multiple imputation integrated to machine learning: Predicting post-stroke recovery of ambulation after intensive inpatient rehabilitation. Scientific Reports, 14(1), 25188. 10.1038/s41598-024-74537-8 [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Forstmeier W., Wagenmakers E. J., Parker T. H. (2017). Detecting and avoiding likely false-positive findings - a practical guide. Biological Reviews, 92(4), 1941–1968. 10.1111/brv.12315 [DOI] [PubMed] [Google Scholar]
  21. Garcia-Rudolph A., Bernabeu M., Cegarra B., Sauri J., Madai V. I., Frey D., Opisso E., Tormos J. M. (2021). Predictive models for independence after stroke rehabilitation: Maugeri external validation and development of a new model. NeuroRehabilitation, 49(3), 415–424. 10.3233/NRE-201619 [DOI] [PubMed] [Google Scholar]
  22. Gelman A. (2017). The “What does not kill my statistical significance makes it stronger” fallacy. Statistical Modeling, Causal Inference, and Social Science. https://statmodeling.stat.columbia.edu/2017/02/06/not-kill-statistical-significance-makes-stronger-fallacy/
  23. Harari Y., O'Brien M. K., Lieber R. L., Jayaraman A. (2020). Inpatient stroke rehabilitation: Prediction of clinical outcomes using a machine-learning approach. Journal of NeuroEngineering and Rehabilitation, 17(1), 71. 10.1186/s12984-020-00704-3 [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Hart T., Ehde D. M. (2015). Defining the treatment targets and active ingredients of rehabilitation: Implications for rehabilitation psychology. Rehabilitation Psychology, 60(2), 126–135. 10.1037/rep0000031 [DOI] [PubMed] [Google Scholar]
  25. Hauser D. J., Ellsworth P. C., Gonzalez R. (2018). Are manipulation checks necessary?. Frontiers in Psychology, 9, 998. 10.3389/fpsyg.2018.00998 [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Hayward K. S., Churilov L., Dalton E. J., Brodtmann A., Campbell B. C. V., Copland D., Dancause N., Godecke E., Hoffmann T. C., Lannin N. A., McDonald M. W., Corbett D., Bernhardt J. (2021). Advancing stroke recovery through improved articulation of nonpharmacological intervention dose. Stroke, 52(2), 761–769. 10.1161/STROKEAHA.120.032496 [DOI] [PubMed] [Google Scholar]
  27. Hayward K. S., Dalton E. J., Campbell B. C. V., Khatri P., Dukelow S. P., Johns H., Walter S., Yogendrakumar V., Pandian J. D., Sacco S., Bernhardt J., Parsons M. W., Saver J. L., Churilov L. (2024). Adaptive trials in stroke: Current use and future directions. Neurology, 103(8), e209876. 10.1212/WNL.0000000000209876 [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Head M. L., Holman L., Lanfear R., Kahn A. T., Jennions M. D. (2015). The extent and consequences of p-hacking in science. PLOS Biology, 13(3), e1002106. 10.1371/journal.pbio.1002106 [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Hoewe J. (2017). Manipulation check. The International Encyclopedia of Communication Research Methods, 12, 1–5. 10.1002/9781118901731.iecrm0135 [DOI] [Google Scholar]
  30. Hordacre B., Austin D., Brown K. E., Graetz L., Parees I., De Trane S., Vallence A. M., Koblar S., Kleinig T., McDonnell M. N., Greenwood R., Ridding M. C., Rothwell J. C. (2021). Evidence for a window of enhanced plasticity in the human motor Cortex following ischemic stroke. Neurorehabilitation and Neural Repair, 35(4), 307–320. 10.1177/1545968321992330 [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Hrobjartsson A., Emanuelsson F., Skou Thomsen A. S., Hilden J., Brorson S. (2014). Bias due to lack of patient blinding in clinical trials. A systematic review of trials randomizing patients to blind and nonblind sub-studies. International Journal of Epidemiology, 43(4), 1272–1283. 10.1093/ije/dyu115 [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Jadad A. R., Moore R. A., Carroll D., Jenkinson C., Reynolds D. J., Gavaghan D. J., McQuay H. J. (1996). Assessing the quality of reports of randomized clinical trials: Is blinding necessary? Controlled Clinical Trials, 17(1), 1–12. 10.1016/0197-2456(95)00134-4 [DOI] [PubMed] [Google Scholar]
  33. Jayasinghe S. A. L., Wang R., Gebara R., Biswas S., Ranganathan R. (2021). Compensatory trunk movements in naturalistic reaching and manipulation tasks in chronic stroke survivors. Journal of Applied Biomechanics, 37(3), 215–223. 10.1123/jab.2020-0090 [DOI] [PubMed] [Google Scholar]
  34. Johnson F., Beeke S., Best W. (2021). Searching for active ingredients in rehabilitation: Applying the taxonomy of behaviour change techniques to a conversation therapy for aphasia. Disability and Rehabilitation, 43(18), 2550–2560. 10.1080/09638288.2019.1703147 [DOI] [PubMed] [Google Scholar]
  35. Karanicolas P. J., Farrokhyar F., Bhandari M. (2010). Blinding: Who, what, when, why, how? Canadian Journal of Surgery, 53(5), 345. [PMC free article] [PubMed] [Google Scholar]
  36. Kim J. K., Lv Z., Park D., Chang M. C. (2022). Practical machine learning model to predict the recovery of motor function in patients with stroke. European Neurology, 85(4), 273–279. 10.1159/000522254 [DOI] [PubMed] [Google Scholar]
  37. Klein O., Hardwicke T. E., Aust F., Breuer J., Danielsson H., Mohr A. H., IJzerman H., Nilsonne G., Vanpaemel W., Frank M. C. (2018). A practical guide for transparency in psychological science. Collabra: Psychology, 4(1), 20. 10.1525/collabra.158 [DOI] [Google Scholar]
  38. Krakauer J. W., Carmichael S. T. (2022). Broken movement: the neurobiology of motor recovery after stroke. MIT Press. [Google Scholar]
  39. Krishnan C., Augenstein T. E., Claflin E. S., Hemsley C. R., Washabaugh E. P., Ranganathan R. (2024). Rest the brain to learn new gait patterns after stroke. Journal of NeuroEngineering and Rehabilitation, 21(1), 192. 10.1186/s12984-024-01494-8 [DOI] [PMC free article] [PubMed] [Google Scholar]
  40. Krishnan C., Ranganathan R., Kantak S. S., Dhaher Y. Y., Rymer W. Z. (2012). Active robotic training improves locomotor function in a stroke survivor. J Neuroeng Rehabil, 9, 57. 10.1186/1743-0003-9-57 [DOI] [PMC free article] [PubMed] [Google Scholar]
  41. Kuch A., Schweighofer N., Finley J. M., McKenzie A., Wen Y., Sanchez N. (2024). Identification of distinct subtypes of post-stroke and neurotypical gait behaviors using neural network analysis of kinematic time series data. bioRxiv. [DOI] [PMC free article] [PubMed]
  42. Kwakkel G., Lannin N. A., Borschmann K., English C., Ali M., Churilov L., Saposnik G., Winstein C., van Wegen E. E. H., Wolf S. L., Krakauer J. W., Bernhardt J. (2017). Standardized measurement of sensorimotor recovery in stroke trials: Consensus-based core recommendations from the stroke recovery and rehabilitation roundtable. Neurorehabilitation and Neural Repair, 31(9), 784–792. 10.1177/1545968317732662 [DOI] [PubMed] [Google Scholar]
  43. Lang C. E. (2024a). Harmonized Upper and Lower Limb Accelerometry Data_Part1 (Version 1) [dataset]. NICHD Data and Specimen Hub.
  44. Lang C. E. (2024b). Lang, Catherine (2024). Harmonized Upper and Lower Limb Accelerometry Data_Part2 (Version 1) [dataset]. NICHD Data and Specimen Hub.
  45. Liew S. L., Lo B. P., Donnelly M. R., Zavaliangos-Petropulu A., Jeong J. N., Barisano G., Hutton A., Simon J. P., Juliano J. M., Suri A., Wang Z., Abdullah A., Kim J., Ard T., Banaj N., Borich M. R., Boyd L. A., Brodtmann A., Buetefisch C. M., Yu C. (2022). A large, curated, open-source stroke neuroimaging dataset to improve lesion segmentation algorithms. Scientific Data, 9(1), 320. 10.1038/s41597-022-01401-7 [DOI] [PMC free article] [PubMed] [Google Scholar]
  46. Lim C. Y., In J. (2019). Randomization in clinical studies. Korean Journal of Anesthesiology, 72(3), 221–232. 10.4097/kja.19049 [DOI] [PMC free article] [PubMed] [Google Scholar]
  47. Lin D. J., Cramer S. C., Boyne P., Khatri P., Krakauer J. W. (2025). High-Dose, high-intensity stroke rehabilitation: Why aren't we giving it? Stroke, 56(5), 1351–1364. 10.1161/STROKEAHA.124.043650 [DOI] [PMC free article] [PubMed] [Google Scholar]
  48. Lohse K. R., Pathania A., Wegman R., Boyd L. A., Lang C. E. (2018). On the reporting of experimental and control therapies in stroke rehabilitation trials: A systematic review. Arch Phys Med Rehabil, 99(7), 1424–1432. 10.1016/j.apmr.2017.12.024 [DOI] [PubMed] [Google Scholar]
  49. Louw S. J. (2002). Research in stroke rehabilitation: Confounding effects of the heterogeneity of stroke, experimental bias and inappropriate outcomes measures. The Journal of Alternative and Complementary Medicine, 8(6), 691–693. 10.1089/10755530260511676 [DOI] [PubMed] [Google Scholar]
  50. Mayernik M. S. (2017). Open data: Accountability and transparency. Big Data & Society, 4(2), 2053951717718853. 10.1177/2053951717718853 [DOI] [Google Scholar]
  51. Miron L., Goncalves R. S., Musen M. A. (2020). Obstacles to the reuse of study metadata in ClinicalTrials.gov. Scientific Data, 7(1), 443. 10.1038/s41597-020-00780-z [DOI] [PMC free article] [PubMed] [Google Scholar]
  52. Munro K. J., Prendergast G. (2019). Encouraging pre-registration of research studies. International Journal of Audiology, 58(3), 123–124. 10.1080/14992027.2019.1574405 [DOI] [PubMed] [Google Scholar]
  53. Murray-Rust P. (2008). Open data in science. Nature Precedings, 34(1), 1–1. 10.1038/npre.2008.1526.1 [DOI] [Google Scholar]
  54. Negrini S., Arienti C., Kiekens C. (2020). Usual care: The big but unmanaged problem of rehabilitation evidence. The Lancet, 395(10221), 337. 10.1016/S0140-6736(19)32553-X [DOI] [PubMed] [Google Scholar]
  55. Nicora G., Pe S., Santangelo G., Billeci L., Aprile I. G., Germanotta M., Bellazzi R., Parimbelli E., Quaglini S. (2025). Systematic review of AI/ML applications in multi-domain robotic rehabilitation: Trends, gaps, and future directions. Journal of NeuroEngineering and Rehabilitation, 22(1), 79. 10.1186/s12984-025-01605-z [DOI] [PMC free article] [PubMed] [Google Scholar]
  56. Norris T. A., Augenstein T. E., Rodriguez K. M., Claflin E. S., Krishnan C. (2024). Shaping corticospinal pathways in virtual reality: Effects of task complexity and sensory feedback during mirror therapy in neurologically intact individuals. Journal of NeuroEngineering and Rehabilitation, 21(1), 154. 10.1186/s12984-024-01454-2 [DOI] [PMC free article] [PubMed] [Google Scholar]
  57. Patel P., Casamento-Moran A., Christou E. A., Lodha N. (2021). Force-Control vs. Strength training: The effect on gait variability in stroke survivors. Front Neurol, 16(12), 12667340. 10.3389/fneur.2021.667340 [DOI] [PMC free article] [PubMed] [Google Scholar]
  58. Quattrocelli S., Russo E. F., Gatta M. T., Filoni S., Pellegrino R., Cangelmi L., Cardone D., Merla A., Perpetuini D. (2024). Integrating Machine Learning with Robotic Rehabilitation May Support Prediction of Recovery of the Upper Limb Motor Function in Stroke Survivors. Brain Sci, 14(8), 759. 10.3390/brainsci14080759 [DOI] [PMC free article] [PubMed] [Google Scholar]
  59. Ranganathan R., Doherty C., Gussert M., Kaplinski E., Koje M., Krishnan C. (2022a). Scientific basis and active ingredients of current therapeutic interventions for stroke rehabilitation. Restor Neurol Neurosci, 40(2), 97–107. 10.3233/RNN-211243 [DOI] [PubMed] [Google Scholar]
  60. Ranganathan R., Lee M. H., Krishnan C. (2022b). Ten guidelines for designing motor learning studies. Brazilian Journal of Motor Behavior, 16(2), 112–133. 10.20338/bjmb.v16i2.283 [DOI] [PMC free article] [PubMed] [Google Scholar]
  61. Riley R. D., Lambert P. C., Abo-Zaid G. (2010). Meta-analysis of individual participant data: rationale, conduct, and reporting. BMJ, 340, 221. 10.1136/bmj.c221 [DOI] [PubMed] [Google Scholar]
  62. Rodgers H., Bosomworth H., van Wijck F., Krebs H. I., Shaw L. (2020). Usual care: The big but unmanaged problem of rehabilitation evidence - Authors’ reply. The Lancet, 395(10221), 337–338. 10.1016/S0140-6736(19)32543-7 [DOI] [PubMed] [Google Scholar]
  63. Ruohonen J., Ollikainen M., Nikouline V., Virtanen J., Ilmoniemi R. J. (2000). Coil design for real and sham transcranial magnetic stimulation. IEEE Transactions on Biomedical Engineering, 47(2), 145–148. 10.1109/10.821731 [DOI] [PubMed] [Google Scholar]
  64. Schepers V. P., Ketelaar M., van de Port I. G., Visser-Meily J. M., Lindeman E. (2007). Comparing contents of functional outcome measures in stroke rehabilitation using the international classification of functioning, disability and health. Disability and Rehabilitation, 29(3), 221–230. 10.1080/09638280600756257 [DOI] [PubMed] [Google Scholar]
  65. Schmidt A. F., Groenwold R. H., Knol M. J., Hoes A. W., Nielen M., Roes K. C., de Boer A., Klungel O. H. (2014). Exploring interaction effects in small samples increases rates of false-positive and false-negative findings: Results from a systematic review and simulation study. Journal of Clinical Epidemiology, 67(7), 821–829. 10.1016/j.jclinepi.2014.02.008 [DOI] [PubMed] [Google Scholar]
  66. Sharp S. A., Brouwer B. J. (1997). Isokinetic strength training of the hemiparetic knee: Effects on function and spasticity. Archives of Physical Medicine and Rehabilitation, 78(11), 1231–1236. 10.1016/S0003-9993(97)90337-3 [DOI] [PubMed] [Google Scholar]
  67. Simmons J. P., Nelson L. D., Simonsohn U. (2011). False-positive psychology: Undisclosed flexibility in data collection and analysis allows presenting anything as significant. Psychological Science, 22(11), 1359–1366. 10.1177/0956797611417632 [DOI] [PubMed] [Google Scholar]
  68. Simmons J. P., Nelson L. D., Simonsohn U. (2021). Pre-registration: Why and how. Journal of Consumer Psychology, 31(1), 151–162. 10.1002/jcpy.1208 [DOI] [Google Scholar]
  69. Smith M. C., Scrivener B. J., Stinear C. M. (2024). Do lower limb motor-evoked potentials predict walking outcomes post-stroke? J Neurol Neurosurg Psychiatry, 95(4), 348–355. 10.1136/jnnp.32.4.348 [DOI] [PubMed] [Google Scholar]
  70. Song F., Hooper L., Loke Y. K. (2013). Publication bias: What is it? How do we measure it? How do we avoid it?. Open Access Journal of Clinical Trials, 8(5), 71–81. 10.2147/OAJCT.S34419 [DOI] [Google Scholar]
  71. Steegen S., Tuerlinckx F., Gelman A., Vanpaemel W. (2016). Increasing transparency through a multiverse analysis. Perspectives on Psychological Science, 11(5), 702–712. 10.1177/1745691616658637 [DOI] [PubMed] [Google Scholar]
  72. Stinear C. M., Byblow W. D., Ackerley S. J., Smith M. C., Borges V. M., Barber P. A. (2017). PREP2: A biomarker-based algorithm for predicting upper limb function after stroke. Annals of Clinical and Translational Neurology, 4(11), 811–820. 10.1002/acn3.488 [DOI] [PMC free article] [PubMed] [Google Scholar]
  73. Stinear C. M., Lang C. E., Zeiler S., Byblow W. D. (2020). Advances and challenges in stroke rehabilitation. The Lancet Neurology, 19(4), 348–360. 10.1016/S1474-4422(19)30415-6 [DOI] [PubMed] [Google Scholar]
  74. Tardy J., Pariente J., Leger A., Dechaumont-Palacin S., Gerdelat A., Guiraud V., Conchou F., Albucher J. F., Marque P., Franceries X., Cognard C., Rascol O., Chollet F., Loubinoux I. (2006). Methylphenidate modulates cerebral post-stroke reorganization. Neuroimage, 33(3), 913–922. 10.1016/j.neuroimage.2006.07.014 [DOI] [PubMed] [Google Scholar]
  75. Tarkka I. M., Kononen M., Pitkanen K., Sivenius J., Mervaalat E. (2008). Alterations in cortical excitability in chronic stroke after constraint-induced movement therapy. Neurological Research, 30(5), 504–510. 10.1179/016164107X252519 [DOI] [PubMed] [Google Scholar]
  76. Tsay J. S., Winstein C. J. (2021). Five features to Look for in early-phase clinical intervention studies. Neurorehabilitation and Neural Repair, 35(1), 3–9. 10.1177/1545968320975439 [DOI] [PMC free article] [PubMed] [Google Scholar]
  77. Tse T., Fain K. M., Zarin D. A. (2018). How to avoid common problems when using ClinicalTrials.gov in research: 10 issues to consider. BMJ, 361, 1452. 10.1136/bmj.k1452 [DOI] [PMC free article] [PubMed] [Google Scholar]
  78. Turner E. H., Mulder R. T., Rucklidge J. J. (2021). Is mandatory prospective trial registration working? An update on the adherence to the international committee of medical journal editors guidelines across five psychiatry journals: 2015-2020. Acta Psychiatrica Scandinavica, 144(5), 510–517. 10.1111/acps.13353 [DOI] [PubMed] [Google Scholar]
  79. Van Stan J. H., Dijkers M. P., Whyte J., Hart T., Turkstra L. S., Zanca J. M., Chen C. (2019). The rehabilitation treatment specification system: Implications for improvements in research design, reporting, replication, and synthesis. Archives of Physical Medicine and Rehabilitation, 100(1), 146–155. 10.1016/j.apmr.2018.09.112 [DOI] [PMC free article] [PubMed] [Google Scholar]
  80. Van Stan J. H., Holmes J., Wengerd L., Juckett L. A., Whyte J., Pinto S. M., Katz L. W., Wolfberg J. (2023). Rehabilitation treatment specification system: Identifying barriers, facilitators, and strategies for implementation in research, education, and clinical care. Archives of Physical Medicine and Rehabilitation, 104(4), 562–568. 10.1016/j.apmr.2022.09.021 [DOI] [PMC free article] [PubMed] [Google Scholar]
  81. Van Stan J. H., Whyte J., Duffy J. R., Barkmeier-Kraemer J. M., Doyle P. B., Gherson S., Kelchner L., Muise J., Petty B., Roy N., Stemple J., Thibeault S., Tolejano C. J. (2021). Rehabilitation treatment specification system: Methodology to identify and describe unique targets and ingredients. Archives of Physical Medicine and Rehabilitation, 102(3), 521–531. 10.1016/j.apmr.2020.09.383 [DOI] [PMC free article] [PubMed] [Google Scholar]
  82. Waddell K. J., Strube M. J., Tabak R. G., Haire-Joshu D., Lang C. E. (2019). Upper limb performance in daily life improves over the first 12 weeks poststroke. Neurorehabilitation and Neural Repair, 33(10), 836–847. 10.1177/1545968319868716 [DOI] [PMC free article] [PubMed] [Google Scholar]
  83. Ward N. (2018). A road map for transforming stroke recovery. Brain, 141(10), 3081–3082. 10.1093/brain/awy248 [DOI] [Google Scholar]
  84. Weiss A., Suzuki T., Bean J., Fielding R. A. (2000). High intensity strength training improves strength and functional performance after stroke. American Journal of Physical Medicine & Rehabilitation, 79(4), 369–376. quiz 391-364. 10.1097/00002060-200007000-00009 [DOI] [PubMed] [Google Scholar]
  85. Weissgerber T. L., Winham S. J., Heinzen E. P., Milin-Lazovic J. S., Garcia-Valencia O., Bukumiric Z., Savic M. D., Garovic V. D., Milic N. M. (2019). Reveal, don't conceal: Transforming data visualization to improve transparency. Circulation, 140(18), 1506–1518. 10.1161/CIRCULATIONAHA.118.037777 [DOI] [PMC free article] [PubMed] [Google Scholar]
  86. Whyte M. (2009). Defining the active ingredients of rehabilitation. Population Health Matters (Formerly Health Policy Newsletter), 22(1), 12. 10.29046/PHM.022.1 [DOI] [Google Scholar]
  87. Winner T. S., Rosenberg M. C., Berman G. J., Kesar T. M., Ting L. H. (2024). Gait signature changes with walking speed are similar among able-bodied young adults despite persistent individual-specific differences. Scientific Reports, 14(1), 19730. 10.1038/s41598-024-70787-8 [DOI] [PMC free article] [PubMed] [Google Scholar]
  88. Winstein C. J., Kay D. B. (2015). Translating the science into practice: Shaping rehabilitation practice to enhance recovery after brain damage. Prog Brain Res, 10(218), 331–360. 10.1016/bs.pbr.2015.01.004 [DOI] [PubMed] [Google Scholar]
  89. Winstein C. J., Stein J., Arena R., Bates B., Cherney L. R., Cramer S. C., Deruyter F., Eng J. J., Fisher B., Harvey R. L., Lang C. E., MacKay-Lyons M., Ottenbacher K. J., Pugh S., Reeves M. J., Richards L. G., Stiers W., Zorowitz R. D., & American Heart Association Stroke Council, C.o.C., Stroke Nursing, C.o.C.C., Council on Quality of, C. & Outcomes, R. (2016). Guidelines for adult stroke rehabilitation and recovery: A guideline for healthcare professionals from the American Heart Association/American stroke association. Stroke, 47(6), e98–e169. 10.1161/STR.0000000000000098 [DOI] [PubMed] [Google Scholar]
  90. Zu W., Huang X., Xu T., Du L., Wang Y., Wang L., Nie W. (2023). Machine learning in predicting outcomes for stroke patients following rehabilitation treatment: A systematic review. PLoS One, 18(6), e0287308. 10.1371/journal.pone.0287308 [DOI] [PMC free article] [PubMed] [Google Scholar]

Articles from Restorative Neurology and Neuroscience are provided here courtesy of SAGE Publications

RESOURCES