The adaptive control consists of a system that could adapt and predict performance in real-time during movements with high accuracy (Hasan and Dhingra, 2021). This type of system can regulate the operation of the interface, from the operating parameters extracted in real-time to reach the best mode of performance (Ljung et al., 2012; Hasan and Dhingra, 2021). Adaptive algorithms can automatically adjust the parameters to categorize individual movement patterns with fewer data and hardware requirements (Chemuturi et al., 2013; Allard et al., 2016).

The user-centered design is one of the requirements of hardware and new technologies development (Manna and Bhaumik, 2013; Gupta et al., 2020). As the current devices are not sufficiently safe to operate physically with people with movement disorders (Chandrasiri et al., 2019), it is required that the system is adaptable to different individuals (Cheung et al., 2017) in order to avoid uncomfortable or unnatural posture, and has multiple degrees of freedom to provide better rehabilitation results (Gupta et al., 2020).

Adaptive algorithms and robot-aided upper limb (UL) rehabilitation must guarantee active training management, patient assistance if necessary (Gupta et al., 2020), and quantify residual muscle strength levels, spasticity, fatigue (Yousif et al., 2019; Bashford et al., 2020), or brain activity (Delijorge et al., 2020). Current studies have shown that exoskeletons can assist in highly repetitive UL task-oriented movement training, and improve daily living activities (Mao and Agrawal, 2011; Sengur et al., 2017; Trigili et al., 2019). Thus, robust methods to automatically classify and identify the UL movement patterns can improve exoskeleton control (Arteaga et al., 2020).

Chu and Patterson (2018) observed that both the design and feedback of many devices need to be improved to maximize patient safety and rehabilitation outcomes. Previous researches (Proietti et al., 2016) have focused on increment this type of device with adaptive algorithms. Moreover, its impact on motor performance needs to be assessed to ensure its usability in individuals with different disorders. Physical (e.g., appearance, weight, and size) and ergonomic features must be considered because they interfere with functionality and affect performance (Merchant et al., 2018).

Successful rehabilitation relies on patient's ability to participate in therapeutic activities and can be influenced by individualized task prescription, considering frequency, duration, and therapy costs. However, passive training is not adequate to recover motor functions. Patient's active training is related to neuroplasticity function (Perez-Ibarra et al., 2018). When devices are properly applied, they can provide better results than conventional approaches, including standardized training, adaptation to loads, and prevention of muscle atrophy (Wu and Chen, 2020).

Smart orthoses based on adaptive algorithms may replicate exercise, and support tools to improve rehabilitation, compliance, and outcomes. From this perspective, we aim to perform the first review related to smart orthoses based on adaptive algorithms and compare usability, algorithms, designs, and benefits between different tools to gather evidence to develop these devices. Therefore, a detailed systematic scoping review will support developers and rehabilitation professionals, through providing answers to the following questions: What mechatronic architectures and characteristics of the systems were used to develop UL orthoses? What adaptive algorithms were used in the orthoses? What tools were used to test orthosis usability? What benefits were found in individuals who have used orthosis?

This paper outlines a scoping review protocol to systematically review the smart UL orthosis based on adaptive algorithms. We will examine the design description, operation, usability test, and registered patent. Eligibility criteria were also defined to include different studies and investigate how UL orthoses are being currently developed. Systematic scoping reviews offer feasible methods for collecting and synthesizing a wide range of evidence (Peters et al., 2015b) and are particularly useful for bringing together evidence from different sources. Information gathered will allow us to correlate published studies with registered patents and propose recommendations to guide new orthoses based on adaptive algorithms.

A scoping review will be designed to identify, analyze gaps, and obtain an overview of the emerging evidence to make clearer what can be investigated in a more appropriate way by a systematic review (Munn et al., 2018). This protocol is based on the York framework proposed by Arksey and O'Malley (2005) that includes five phases: identifying the research question (1), identifying relevant studies (2), study selection (3), charting the data (4), and collating, summarizing and reporting the results (5).

In addition, the Joanna Briggs Institute (JBI) (Peters et al., 2015a) framework will be used to clarify and provide a greater relationship between the title, review objective(s), question(s), and inclusion criteria. Thus, a field-specific structure based on an initial exploration of studies and aspects related to orthosis description, device operation, control system characteristics, acquisition algorithm and its processing techniques, adaptive algorithm parameters, characterization of the participants, intervention description, and results of each study, will guide each phase of the review.

The present protocol is reported according to the Preferred Reporting Items for Systematic reviews and Meta-Analyses (PRISMA-P) protocols (Moher et al., 2015) as suggested in the PRISMA Extension for Scoping Review (PRISMA-ScR) guidelines (Tricco et al., 2018) (see checklist in Supplementary File 1), and its updated checklist (Rethlefsen et al., 2021) (see updated checklist in Supplementary File 2).

This review will explore specific questions such as feasibility, orthosis effectiveness, and orthosis characteristics and concepts (Munn et al., 2018). Therefore, the review questions were previously defined to provide the roadmap for subsequent stages.

During the consultation with our systematic scoping review experts (EM, ARRL, RAMV, and DAPN), a set of questions were elaborated for each of the following aspects (Table 1): (1) description of both the orthosis (2) and the adaptive algorithms used, (3) tools used to test usability, and (4) benefits for the general population.

It is worth mentioning that the research questions of this study will not be limited to those presented in Table 1. Further questions can be discussed based on data analyses at the systematic scoping review elaboration.

The JBI suggests using the Population, Concept, and Context (PCC) to construct scope review questions since it is a less restrictive alternative to the PICO (Peters et al., 2015a).

The target “population” will include people aged ≥18 years, regardless of gender, and health status (i.e., healthy or unhealthy with acute or chronic diseases). The “concept” will cover all studies that developed smart orthoses based on adaptive algorithms. The “context” will cover comparisons among mechatronic architecture, algorithms, and types of human-machine interface controller. We will also compare the impact of UL orthosis controlled by adaptive algorithms on motor function and physical therapy treatment.

Systematic scoping reviews provide a large area of scientific evidence on a particular topic. The search strategy elaboration using keywords and synonyms based on the Medical Subject Headings (MeSH) must also be broad and will consider four significant areas of the theme: body segment, orthosis, physical rehabilitation, and adaptive algorithms. Two sections will be created to describe the steps for selecting the relevant studies: search resources and search strings.

The inclusion criteria for selecting the articles will be:

On the other hand, the criteria for selecting the patents are only about the types of orthoses similar to the one described above, and the following:

We chose articles and patents published in several languages in the last 10 years to obtain an overview of the evidence about the adaptive algorithms used as a control strategy for UL orthoses. In this way, we will be able to guide developers of this type of orthosis and professionals who use it in rehabilitation.

Five authors will construct a descriptive summary of the results: LJH and APMF will insert the article data, while APMF, JAA, and SPNN will insert the patent data. For this stage, a data charting form will be developed with variables corresponding to the research questions. The components of the identified frameworks will be used to determine an initial set of variables updated as “articles being reviewed.” A description of frameworks and their components is provided below and summarized in Table 3.

The extracted data will be inserted in a table in which rows will represent the included articles and patents, columns will represent the variables, and cells will contain the strategies used for the relevant variables. To analyze this data set, frequency and trend analyses will be used. The risk of bias assessment and confidence in cumulative evidence will also be performed.

Trend analysis will be used to present the research evolution based on several variables and associations. The organization by clusters will combine the studies for data exploration (e.g., explore the performance of a specific machine learning technique and obtain specific findings), resulting in a map of studies represented as a bubble plot, graph, or table.

In this stage, two main aspects will be considered: (1) the demographic geographic characteristics in which the clustering will be used to define the patterns of the state of the art development and (2) correlations between the different centers in the world. Concomitantly, a regression analysis will be carried out to explore patterns concerning the development of the topic over time.