Adaptive Algorithms as Control Strategies of Smart Upper Limb Orthosis: A Protocol for a Systematic Scoping Review

Abstract

Introduction: Adaptive algorithms for controlling orthosis emerged to overcome significant problems with automatic biosignal classification and personalized rehabilitation. Smart orthoses are evolving fast and need a better human-machine interaction performance since biosignals, feedback, and motor control dynamically change and must be adaptive. This manuscript outlines a scoping review protocol to systematically review the smart upper limb (UL) orthoses based on adaptive algorithms and feasibility tests.Materials and Methods: This protocol was developed based on the York framework. A field-specific structure was defined to achieve each phase. Eleven scientific databases (PubMed, Web of Science, SciELO, Koreamed, Jstage, AMED, CENTRAL, PEDro, IEEE, Scopus, and Arxiv) and five patent databases (Patentscope, Patentlens, Google Patents, Kripis, J-platpat) were searched. The developed framework will extract data (i.e., orthosis description, adaptive algorithms, tools used in the usability test, and benefits to the general population) from the selected studies using a rigorous approach. Data will be described quantitatively using frequency and trend analysis methods. Heterogeneity between the included studies will be assessed using the Chi-test and I-statistic. The risk of bias will be summarized using the latest Prediction Model Study Risk of Bias Assessment Tool.Discussion: This review will identify, map, and synthesize the advances about the description of adaptive algorithms for control strategies of smart UL orthosis using data extracted from patents and articles.

Machine learning as a control strategy for smart upper limb orthosis: a scoping review

“…To ensure transparent findings of the review of this research topic, we delineated and performed according to the Preferred Reporting Items for Systematic reviews and Meta-Analyses (PRISMA-P) protocols [18] as suggested in the PRISMA Extension for Scoping Review (PRISMA-ScR) guidelines [19] (see checklist in Additional File 1), and its updated checklist [20] (see updated version in Additional File 2). Additionally, we published a protocol for our scoping review [21] to strengthen both results and quality of this study. This section provides an overview of the stages.…”

“…All the screening process was conducted and reported according to the PRISMA framework [22,23] using Rayyan (http://rayyan.qcri.org) [24]. Two authors of the research team (LJH, APMF) examined a random sample of 10% of the found articles based on the predefined inclusion and exclusion criteria [21]. The titles and abstracts of the search results were independently screened by four authors (LJH, APMF, JAA, SPNN) and full-text articles were obtained for relevant studies.…”

“…The variables about each aspect were determined according to the research questions and the underlying frameworks (a list of variables is provided in the results) [21]. The data extraction was carried out in a discussion by all authors, simultaneously with data tabulation, to ensure that the range of values for each variable was accurately captured.…”

“…The search strategy was performed regarding the following four categories: body segment (hand, elbow, wrist, shoulder, arm, fingers, upper extremity, upper limb), orthosis (orthosis, orthoses, exoskeleton, orthotic device, orthosis device, orthoses device, robotic device, robotic, wearable robot, exosuit, wearable orthoses, wearable orthosis, wearable assistive robots, wearable exosuit), physical rehabilitation (physical rehabilitation, motor rehabilitation, physical medicine, telerehabilitation), and ML algorithms (ML, supervised ML, unsupervised ML, semi-supervised ML, reinforcement learning, artificial intelligence, adaptive algorithms, neural network (NN), artificial NN, k-nearest neighbors (KNN), linear regression, logistic regression, support vector machines (SVM), decision trees, random forests (RF), extreme gradient boosting, k-means, hierarchical cluster analysis, expectation maximization, principal component analysis, kernel principal component analysis, locally-linear embedding, T-distributed stochastic neighbor embedding, Q-learning, state-action-reward-stateaction, deep Q network, deep deterministic policy gradient, computational intelligence, machine intelligence, computer reasoning, computer vision systems, knowledge acquisition, knowledge representation, logic fuzzy, fuzzy control, deep learning) for both databases and patents, according to our previous study [21]. All database searches were done without restriction to the language and year of publication.…”

“…To be included, articles had to: (1) include participants over 18 years old, either healthy or with the clinical diagnosis of any acute or chronic UL movement disorder with any degree of severity; (2) contain the development of UL orthoses based on ML algorithms and its usability test has been performed or not; (3) report instrumented and noninstrumented measurements used as outcome measures for motion analysis before and during use of the device; (4) describe the physical and control characteristics of UL orthosis; (5) be full-text studies and experimental study designs, including randomized controlled trials (RCTs). Non-RCTs and descriptive epidemiological study designs were included due to the absence of RCTs as pre-established in our protocol [21]. All database searches were without restriction to the language and year of publication.…”