Purpose Virtual reality (VR) games are increasingly being explored as tools for physical and mental rehabilitation, yet their cognitive and psychological demands relative to conventional 2D games remain insufficiently understood. In rehabilitation-oriented tasks, factors such as mental workload (MWL), situational awareness (SA) and affect play a critical role in adherence and safety. As an initial, pre-clinical step, this study examines how these demands differ between immersive VR and traditional 2D gameplay. Design/methodology/approach A within-subjects study was conducted with 21 healthy adults who performed game-based tasks of varying difficulty in both VR and 2D environments. MWL, affect and SA were assessed after each condition using self-assessment questionnaires. Findings Compared with the 2D environment, VR resulted in significantly higher overall MWL, effort and physical demand, particularly at the easier task difficulty. VR was also associated with higher negative affect at the easy level; however, negative affect did not further increase at the high level, while positive affect remained stable across difficulty levels. Originality/value The findings offer foundational insight into the cognitive and affective demands associated with immersive VR gameplay prior to clinical application. The results underscore the need to balance workload, task difficulty and emotional intensity in the design of rehabilitation-oriented VR games, especially in scenarios where exercises are inherently challenging and sustained engagement is essential.
Purpose This scoping review maps the extent, range and nature of evidence on how artificial intelligence (AI)–powered assistive technologies support disabled students in higher education, identifying the types of tools in use, the disability groups served, the benefits and barriers reported, and the priority research gaps that emerge from the evidence. Design/methodology/approach Following the PRISMA-ScR framework (Peters et al., 2020), seven academic databases were searched for peer-reviewed studies published between January 2020 and February 2026. Search terms were developed iteratively using Medical Subject Headings, the Education Resources Information Center (ERIC) thesaurus and the ACM Computing Classification System, with pilot searches to confirm recall. From 1,847 initial records, 47 studies met all inclusion criteria after title/abstract and full-text screening. Findings Six AI tool categories are identified and mapped to disability groups and study counts: large language models (n = 14), speech and language processing (n = 11), intelligent tutoring systems (n = 9), computer vision tools (n = 7), adaptive learning platforms (n = 4) and affective/behaviour recognition systems (n = 2). Benefits include personalised support, reduced cognitive load, improved writing and psychosocial gains. Barriers include algorithmic bias, cost-driven digital divides, institutional policy vacuums and exclusion of disabled students from AI design. Seven priority research gaps emerge from the review findings. Research limitations/implications The review is restricted to peer-reviewed, English-language publications, which limits conclusions about research in non-English-speaking contexts and excludes grey literature. The rapidly evolving pace of AI development means some findings may be superseded. These limitations are discussed in a dedicated Limitations section. Practical implications Universities should develop disability-inclusive AI policies that explicitly recognise generative AI as a legitimate form of assistive technology, ensure equitable access regardless of cost and integrate AI tools within universal design for learning frameworks. Social implications AI holds significant potential to narrow the persistent graduation gap between disabled and non-disabled university students. Realising this requires inclusive design, equitable access and genuine participation by disabled students in shaping the technologies that affect them. Originality/value This is the first scoping review in the Journal of Enabling Technologies comprehensively mapping AI-powered assistive technologies for disabled students across all disability categories within higher education. It synthesises the most recent 2024–2026 empirical evidence, presents a PRISMA-ScR flowchart and proposes a structured seven-gap research agenda grounded in the review findings.
Purpose The aim of this article is to analyze the role of Socially Assistive Robotics (SAR) as an enabling technology for the inclusion of students with autism spectrum disorder (ASD) in school settings. The study seeks to bridge the gap between research potential and practical application by evaluating the current methodological landscape, the ecological validity of research designs, and the transferability of these technologies to real classroom contexts. Design/methodology/approach A systematic literature review was conducted in January 2026 using the Scopus and Web of Science databases. A total of 33 primary research articles published between 2020 and 2025 were identified, all of which employed robotic platforms with children or adolescents with ASD in educational contexts. The analysis applies a thematic synthesis within a descriptive analytical framework to categorize methodological characteristics, technological platforms, and the degree of pedagogical integration. Findings The findings confirm that SAR functions as a catalyst for inclusion through three mechanisms: as a social bridge, a structured predictor, and a personalized motivator. However, the review reveals significant technological fragmentation (29 different platforms, with the NAO robot leading at 30.3%) and a lack of reporting on practical implementation. Only 9.1% of studies provide cost-related information, and the majority of interventions (57.6%) consist of isolated sessions not integrated into the curriculum. Critical barriers are identified, including limited robot autonomy and the need for specialized teacher training. Practical implications The study highlights the need to adopt translational study designs and co-design protocols with educators to ensure sustainability. The implications suggest that, for effective inclusion, robotics should be embedded within individualized education plans (IEPs) and daily classroom routines, moving beyond the conception of robots as isolated therapeutic tools. The DivInTech project is proposed as a collaborative ecosystem model that promotes training and open resource sharing to overcome cost and scalability barriers. Originality/value Unlike previous reviews that focus primarily on clinical outcomes, this study adopts a critical methodological lens. Its originality lies in examining how research is designed and reported, identifying tensions between experimental control and school feasibility. Furthermore, the article proposes a standardized methodological framework and concrete actions to enhance the rigor and real-world impact of future interventions.
Purpose This study aimed to assess the user experience (UX) of South African parents of children with disability (CWD) using the parent network (PN) and let's talk parents (LTP) platforms, and to examine associations between demographic factors and UX Honeycomb dimensions. It includes a comparative analysis of demographics in relation to the UX honeycomb subsections. Design/methodology/approach A quantitative, cross-sectional survey was done. Using total population sampling, 359 participants completed an online survey based on Morville's UX Honeycomb model, integrating validated instruments (MAUX-C and SUPR-Qm). Responses were analysed through descriptive and inferential statistics, including linear regression analyses to examine associations between demographical variables and UX dimensions. The study ensured accessibility with multilingual translation and ethical rigor through voluntary participation, informed consent and anonymised data collection. While reliability testing was not conducted, face and content validity were supported. Findings offer insights into UX dimensions relevant to digital inclusion and participatory engagement in disability-focused networks. Findings Most participants were mothers aged 31–40, with children predominantly aged 6–10, diagnosed mainly with cerebral palsy or autism. The findings highlight strong user satisfaction, particularly in usefulness, desirability, accessibility and credibility, with value rated highest (90%). However, findability emerged as a challenge, with variability in users' ability to locate specific content. Inferential analysis showed that age was significantly associated with usability, usefulness, and desirability, although effect sizes were small. Male participants reported lower scores for usefulness, accessibility, credibility, and value. These findings should be interpreted with caution due to the small number of male participants. Widowed participants reported lower findability, and participants using private healthcare reported lower accessibility. Overall, demographic variables explained only a small proportion of variation in UX. Research limitations/implications The sample was predominantly female and drawn from an existing digital network, which may limit generalisability. However, caregiving responsibilities are predominantly undertaken by women, a pattern that is particularly pronounced in the Global South, including South Africa (Latulippe et al., 2017; Budlender, 2021). The survey incorporated adapted instruments without internal consistency testing, and the cross-sectional design limits insight into long-term engagement and understanding of sustained impacts. The potential for using the collected data in advocacy efforts is also somewhat constrained. It is a limitation of the PN and LTP to have focused heavily on the children and not parents who have disabilities, and as such next iterations of these tools should take that into account. The researcher's dual role as both insider and outsider was acknowledged, but it may still be viewed as a limitation, potentially influencing the study's objectivity and interpretation. Practical implications This study highlights the transformative role of UX in disability-focused digital platforms, showing that PN and LTP foster identity affirmation, peer support and civic engagement. While usefulness, desirability and empowerment received strong endorsements, gaps in findability and inclusivity, especially for male and widowed caregivers, suggest the need for targeted design improvements. Findings further indicate that improvements in navigation, accessibility and inclusive design may enhance engagement across diverse user groups. Future iterations should prioritise co-design with underrepresented users, enhanced accessibility, and gender-responsive navigation strategies. By embedding participatory design principles, digital tools can amplify marginalised voices, democratise information access and drive systemic advocacy, particularly in the Global South, where equity-focused UX is key to sustained engagement and impact. Social implications As addressed in findings, methodology and practical implications. Originality/value This study underscores the transformative role of UX in disability-focused digital platforms, showing that PN and LTP foster identity affirmation, peer support and civic engagement beyond mere functionality. While usefulness, desirability and empowerment were highly rated, gaps in findability and inclusivity, specially for male and widowed caregivers, highlight areas for targeted design improvements. This study contributes to limited quantitative evidence on UX of disability-focused caregiver platforms in the Global South and demonstrates how UX relates to caregiver participation and voice. By embedding participatory design principles, digital tools can amplify marginalised voices, democratise information access and drive systemic advocacy, particularly in the Global South, where UX must center dignity, engagement and lasting impact.
PurposeThis scoping review explores the landscape of assistive technology (AT) training for professionals working in special education contexts. Despite the recognised importance of AT in supporting equitable access to learning, the professional preparation and in-service training of teachers and other learner-facing education staff remains inconsistent. Design/methodology/approachFollowing the PRISMA-ScR framework, a systematic search identified sources exploring or discussing AT training, professional perceptions, and reported outcomes. Findings were mapped across themes including training content, delivery, accessibility and effectiveness. FindingsThe review highlights that AT training is highly valued by professionals but inconsistently available, with systemic barriers to access and variable recognition of attainment. Pre-qualification training in AT is limited and variable across disciplines, while post-qualification learning or continuing professional development (CPD) is often self-directed or locally developed. Barriers to training include time, funding and lack of awareness of available opportunities. Flexible and blended models, incorporating online and hands-on components, may be effective AT training approaches. Confidence and self-efficacy emerged as the most frequently measured indicators of training success, with several studies reporting gains in these areas following structured training. However, outcome measurement remains inconsistent, and few validated instruments exist to assess AT training effectiveness. Originality/valueAcross the literature, professionals expressed strong motivation to improve their AT competence but highlighted systemic gaps in formal education, accreditation and expert support. Notably, despite the importance attached by practitioners to training learning support staff, this group may find it harder than others to access training opportunities. Addressing these training gaps is essential to ensure equitable implementation of AT in education. The review identifies priorities for future research, including the development of standardised outcome measures and models for scalable, accredited professional learning in AT.
PurposeThis study explored the needs, preferences, and pathways for integrating Artificial Intelligence (AI) into assistive technologies (AT) among persons with disabilities (PWDs) in Ghana.Design/methodology/approachThis study used a mixed-method approach to examine the needs and preferences of users for integrating AI Solutions into Existing Assistive Technology Innovations in Ghana. The study used a structured questionnaire and interview guide to collect data from 385 respondents with mobility, hearing, and visual impairments across three regions: Greater Accra, Ashanti, and Northern.FindingsThe findings revealed that affordability, ease of use, and customisation were the most significant determinants of AI-enabled AT adoption, while voice interaction emerged as the most preferred interface across disability types. Despite growing awareness, adoption remains hindered by infrastructural deficits, limited internet access, high device costs, and low digital literacy. The results highlight a growing demand for affordable, context-specific, and adaptive technologies that align with Ghana's socio-economic realities.Research limitations/implicationsA key limitation of this study is its reliance on quantitative self-reported data, which may not fully capture the depth and complexity of user experiences with AI-powered assistive technologies. Respondents might have over- or under-reported their awareness, preferences, or access due to recall bias or social desirability tendencies, especially when discussing technology-related issues. Moreover, the study's cross-sectional nature restricts its ability to assess changes in perceptions and usage patterns over time, particularly as digital literacy and AI exposure continue to evolve in Ghana. Although the inclusion of three regions - Greater Accra, Ashanti, and Northern - provides valuable socio-economic diversity, the findings may not fully represent persons with disabilities in more remote or under-resourced areas, where infrastructural challenges and socio-cultural barriers may differ significantly.Practical implicationsThe study findings demonstrate the need for interdisciplinary research focusing on adaptive and multimodal interface design, recognising that disability groups have distinct usability requirements beyond universal design assumptions. The results also imply that there is a need to institutionalise participatory governance frameworks that actively involve PWDs in AI-AT design, regulation, and evaluation processes to ensure technologies align with real user needs.Social implicationsImproved affordability and usability of AI-AT can significantly enhance social inclusion, independence, and participation of PWDs in education, employment, and community life.Originality/valueThe study concludes that inclusive, user-centred AI innovation, supported by strong policy frameworks and investment in digital infrastructure, is crucial for achieving equitable access to assistive technologies and advancing disability inclusion in Ghana.
Purpose Augmentative and alternative communication (AAC) systems support individuals who cannot rely on natural speech by enabling communication through symbol selection on electronic devices. Effective AAC use requires motor learning, yet limited research has examined how practice structure influences AAC performance. This feasibility study examined whether different practice schedules: blocked practice (repeated practice of the same target symbol before moving on to the next target) versus random practice (practice of target symbols in a randomized, nonconsecutive order) are associated with differences in response time and accuracy retention during AAC symbol selection. Design/methodology/approach Eight young adults were randomly assigned to either a blocked or random practice schedule and completed four days of symbol-selection practice on a high-technology AAC device. Participants practiced selecting target vocabulary items arranged within a grid-based display. Response time and accuracy were measured immediately after training and again one week later to assess short-term retention. Mixed-model analyses were used to evaluate whether the experimental paradigm was sensitive to differences in performance associated with practice schedule, retention interval and task demands. Findings Accuracy remained consistently high across both practice schedules, with no detectable differences between conditions, suggesting a ceiling effect for this outcome measure. Response time showed greater variability: random practice was associated with faster performance during acquisition but slower response times at the one-week retention test, whereas blocked practice demonstrated more stable response times across sessions. These patterns indicate that response time may be more sensitive than accuracy to practice-schedule manipulation within this task design. Originality/value This study is the first to examine the feasibility of comparing blocked and random practice schedules in high-technology AAC symbol training. The findings provide preliminary insight into how practice structure may influence motor performance and inform the design of future AAC motor learning studies incorporating more complex tasks, longer training periods and AAC user populations.
Purpose - This study evaluates whether a custom-built Virtual Reality (VR) Serious Game - SAVIRE - can serve as an enabling technology to overcome barriers in traditional exposure therapy for adolescents with social anxiety disorder (SAD). It specifically examines the system's usability and clinical efficacy in a resource-constrained public health context. Design/methodology/approach - A randomized pilot trial (N = 20) was conducted using a mixed-methods approach. Participants were assigned to either standard CBT or VR-supported CBT. Efficacy was measured via LSAS and SPIN scales, alongside MEEGA + usability metrics and objective in-game behavioral tracking. Findings - The VR-supported intervention yielded significantly greater symptom reduction than standard care for both LSAS (p < 0.001, eta(2 )= 0.75) and SPIN (p = 0.004, eta(2) = 0.40). Regression analysis confirmed that in-game engagement significantly predicted clinical gains (p = 0.047). Usability testing (MEEGA+) verified high accessibility and relevance for the target demographic. Research limitations/implications - The study is limited by its small sample size (N = 20) and single-site context. As a pilot trial, these results provide preliminary validation, but larger multi-site studies are required to confirm broad scalability. Practical implications - The system offers an objective clinical tracking tool and a cost-effective implementation model via consumer-grade hardware, effectively overcoming logistical barriers in public health settings. Social implications - By aligning with the digital habits of adolescents, the intervention reduces stigma and demonstrates how immersive technology can bridge the digital divide in regions with mental health professional shortages. Originality/value - This research validates a framework for operationalizing complex therapeutic tasks using Social Learning Theory and Bloom's Revised Taxonomy into a user-friendly format, enabling standardized care delivery in resource-limited settings.
PurposeSelf-care (mHealth) apps play an important role in chronic disease management by enabling patients to track medications, monitor routines and adopt healthier lifestyles. In Fiji, where diabetes prevalence is among the highest globally, such apps could enhance self-management. However, the effectiveness of these tools depends significantly on usability during the onboarding process. Given that smartphones and tablets differ in screen size and input methods, user performance on these devices may vary.Design/methodology/approachThis study evaluated the onboarding usability of self-care apps among diabetes patients in Fiji. A task-based usability evaluation was conducted with first-time users on both smartphones and tablets.FindingsFindings showed that although most participants completed the assigned tasks, smartphones presented greater usability challenges. These included higher error rates, longer task completion times and lower satisfaction levels compared to tablets.Originality/valueThe results emphasize the importance of device-specific design considerations for mHealth applications. These insights are valuable to app developers and health policymakers in Fiji seeking to improve the adoption and effectiveness of self-care apps for diabetes management.
Purpose Dyslexia poses significant challenges to digital reading, often resulting in increased cognitive load, reduced comprehension, and limited engagement with complex texts. While assistive technologies exist, many fail to dynamically adapt to user needs or address cognitive processing categories effectively. This study aimed to design and evaluate an AI-based cognitive load reduction (CLR) Tool to enhance digital reading accessibility, reduce perceived mental effort, and improve perceived comprehension for dyslexic learners and professionals. Design/methodology/approach Guided by cognitive load theory, the CLR Tool integrates: (1) summarisation and complexity scoring to reduce intrinsic load, (2) text-to-speech, translation, and sentiment analysis to reduce extraneous load and (3) keyword highlighting and note-taking to support germane load. The platform was built using React.js, Node.js, Firebase and OpenAI APIs. A mixed-methods evaluation was conducted with dyslexic and non-dyslexic participants through pre- and post-intervention surveys, usability testing and technical validation. Findings In a pilot mixed-methods evaluation, participants reported high usability and indicative benefits in perceived comprehension and reduced perceived strain when reading complex digital text. Outcomes were primarily captured via low-burden binary self-report items, supplemented by qualitative feedback and basic usage analytics. Findings provide preliminary support that a CLT-informed, integrated reading workflow may improve perceived comprehension and reduce perceived mental effort for dyslexic users. Originality/value The CLR Tool represents a scalable framework for inclusive, adaptive educational technologies that can extend benefits beyond the target population. Its originality lies in combining cognitive load theory with AI-enabled summarisation, complexity scoring, text-to-speech, translation, sentiment analysis, keyword highlighting and note-taking within a single integrated reading-support workflow.
PurposeOngoing shortages of mental health professionals in the USA underscore the need for scalable, technology-mediated supports. Such supports should not only extend access but also preserve core therapeutic ingredients, including human-like conversational scaffolding via AI and immersive, therapy-like contexts via VR. This paper reports on the design and formative pilot evaluation of a low-fidelity prototype for a conceptual AI-based, VR-supported mental health application. Design/methodology/approachBecause the goal was to probe usability and perceptions at an early stage, we used low-fidelity materials to simulate the VR therapy flow and conceptual AI-assisted interactions rather than implementing full AI or VR functionality. Ten young adults with mild to moderately severe depression interacted with the low-fidelity prototype, completed the Usefulness, Satisfaction, and Ease of use questionnaire, and joined semi-structured interviews. FindingsQuantitative ratings indicated that the prototype was learnable and generally acceptable for this population. Thematic analysis of the interviews yielded two broad areas of concern – (1) current mental health care barriers and (2) expectations for AI- and VR-supported therapeutic experiences – which pointed to design needs such as anonymity, culturally sensitive options, lightweight onboarding, and more human-like timing and nonverbal cues. Taken together, these findings suggest that the proposed AI-supported VR concept is understandable and worth further digital prototyping, but future work must move beyond low-fidelity simulations to evaluate real interactions and emotional engagement. Originality/valueThis study demonstrates the feasibility of AI and VR-supported mental health interventions and provides early insights to guide the design of engaging, and accessible digital therapy.
PurposeOur study's purpose is to understand the possibilities and challenges of using 3D printed tactile maps to improve campus navigational services for blind and low-vision (BLV) students.Design/methodology/approachWe used a participatory design (PD) approach with co-design sessions, focus groups and interviews with students with disabilities and university accessibility service providers. In this study, we focus on findings from a focus group with expert accessibility service providers.FindingsKey contributions include insight into the navigational challenges and preferences of BLV students in a university environment. We further discuss the importance of collaboration between researchers and university accessibility service providers and the need to adopt accessible principles at an institutional level to support do-it-yourself (DIY) assistive technology approaches, increasing access in formal learning contexts.Research limitations/implicationsThe scope of our participant pool was limited, involving a small number of students from a single US institution. Also, the inaccessibility of the 3D design tools we had access to prevented our participants from engaging directly in the map design process.Practical implicationsOur work contributes to the accessibility and assistive technologies research communities by highlighting the challenges faced by BLV students when navigating their campus and by demonstrating the efficacy of PD practices in developing 3D-printed tactile maps that elevate their needs and open up possibilities for advocacy and multi-stakeholder conversation.Originality/valueOur study shows the perspectives of university accessibility service providers on the possibility of using DIY assistive technology, such as a 3D-printed tactile campus map, to facilitate accessibility services for BLV university students in the United States.
PurposeThis opinion paper aims to examine how artificial intelligence (AI), including generative AI, is reshaping education for people with disabilities. Traditional classroom environments have long created barriers to equitable learning, but emerging AI-driven tools now offer adaptive, personalized support that enhances accessibility, engagement and independence. The paper explores how AI functions as a socio-technical infrastructure that reconfigures disability and ability in educational contexts. By synthesizing evidence across AI in education, assistive technologies and inclusive pedagogy, it highlights three interrelated themes – personalization, barrier reduction and democratized access – to illustrate how AI can enable more inclusive and responsive learning environments. Design/methodology/approachAs an opinion piece, this paper synthesizes existing literature to examine current and emerging uses of AI in inclusive education. It integrates research on intelligent tutoring systems, educational data mining, universal design for learning and AI-powered assistive technologies. The approach emphasizes conceptual analysis rather than empirical evaluation, drawing on illustrative examples to demonstrate how personalization, generative AI content creation, real-time accessibility features and cloud-based tools support diverse learners. The synthesis is organized around themes that highlight both opportunities and systemic considerations involved in integrating AI into disability-inclusive educational practice. FindingsThe synthesis indicates that AI has substantial potential to reduce long-standing educational barriers by enabling real-time adaptation to individual needs, generating accessible learning materials and embedding assistive features into mainstream technologies. Intelligent tutoring systems, speech-to-text tools, generative AI applications and cloud-based platforms collectively enhance personalization, accessibility and learner engagement. However, the review also identifies critical challenges, including privacy concerns, algorithmic bias affecting disabled learners, limited educator preparedness and questions around governance and equity. These issues must be addressed to ensure that AI-driven innovations advance inclusive rather than exclusionary educational practices. Research limitations/implicationsThe findings carry important implications for practice, policy and research. Educators can use AI to identify barriers, personalize support and scaffold learners' use of assistive tools, but they require targeted professional development. Institutions and policymakers must prioritize accessible procurement, transparent governance and infrastructure that protects disability-related data. Future research should evaluate not only learning outcomes but also participation, autonomy and equity for diverse disabled learners. Co-design with disabled students and educators will be essential for shaping responsible and inclusive AI deployment. Together, these implications highlight the need for a coordinated approach to building accessible, AI-enabled learning environments. Originality/valueThis paper offers a novel conceptual framing by positioning AI, particularly generative AI, as a socio-technical infrastructure that reshapes how disability and ability are configured in educational settings. It synthesizes current evidence through three integrative themes: personalization, barrier reduction and democratized access. The paper foregrounds disability-specific considerations often absent from mainstream AI-in-education debates and critically examines both opportunities and risks. By articulating clear implications for educators, institutions and policymakers, it provides a timely and distinctive contribution to ongoing discussions about inclusive AI adoption in education.
PurposeChildren undergoing Magnetic Resonance Imaging (MRI) often experience anticipatory anxiety, which can lead to distress, rescheduled scans or sedation. This study evaluated the acceptability and impact of Xploro, a digital interactive app designed to improve children's understanding of MRI procedures and reduce anxiety.Design/methodology/approachThe GLOWING STARS study used a before-and-after design. Children aged 4-11 attending diagnostic MRI scans in secondary care paediatric NHS Trusts accessed Xploro one-to-one on the day of their appointment. Children and parents/caregivers completed paper proformas assessing perceived knowledge and emotional responses before and after using the app. Qualitative data were analysed thematically, and quantitative data summarised descriptively.FindingsThe study indicated that Xploro was highly acceptable and effective in enhancing procedural understanding, improving preparation and increasing emotional security before MRI scans, especially for children aged 7 to 11. This study is the first to involve children aged 4-6 years using the Xploro digital interactive intervention for MRI preparation, and the first to deliver Xploro on-site at the ANONYMISED hospital NHS Trust. This study provides valuable evidence guiding recommendations for the development of digital interventions for children under 7 years old (similar to 3-6).Research limitations/implicationsDue to the very small sample size of the data, caution should be taken when interpreting results about app-based strategies alongside relational support. Further research with a larger sample and with more balanced representation across diagnostic categories is needed to clarify the child-digital technology relationship and explore the role of adult facilitation in digital health app engagement among children (7-11 years old). Although the subgroup sizes were too small to draw reliable conclusions about children with physical disabilities and neurodevelopmental conditions, the qualitative data from the 4 children with conditions suggest a potential disparity in the perceived effectiveness of the app for children with physical disabilities and neurodiverse developmental needs.Originality/valueThis study is the first to involve children aged 4-6 years using the Xploro digital interactive intervention for MRI preparation, and the first to deliver Xploro on-site at the ANONYMISED hospital NHS Trust. This study provides valuable evidence guiding recommendations for the development of digital interventions for children under 7 years old (similar to 3-6).
PurposeThis study examines the usability, acceptability and technical implementation of an adaptive digital platform that supports executive functions and emotion regulation in children with specific educational support needs (SESN), aiming to facilitate their personal, social, and academic growth.Design/methodology/approachA process of co-design was followed for the definition of the functionality of the digital platform, including the participation of focus groups of teachers working with children with SESN and the participation of independent panels of neurodivergent adults. The digital platform for assessing and training executive functions has been developed, integrating gamification, artificial intelligence and sensory technologies (eye-tracking and wearables). A total of 27 children with SESN aged 8-14 years participated in the pilot study.FindingsResults suggest that +85% of children mention appropriate game time and ease of navigation. Approximately half of the teachers (48%) report that the game settings are excellent to very good. These contributions provide guidelines for IT technicians and constitute standards of use for teachers and children with NDDs. This study has demonstrated the functionality, safety and usability of the developed digital platform. It also helped in the identification of opportunities for improvement in the design and implementation of that platform, pushing it forward for subsequent experimental validation.Originality/valuePrevious studies investigated the effectiveness of separate cognitive tasks for training executive functions with children with SESN. However, no study has been conducted to validate a digital platform as a unitary assessment battery for these functions and its feasibility to be used in a school setting. This digital platform is also the first one to integrate gamification, artificial intelligence and sensory technologies for assessing and training executive functions in children with SESN.
PurposeThis paper addresses the research gap regarding ways automatic speech recognition can support AI in learning beyond just becoming a useful tool for the transcription of lectures for students with accuracy rates in English improving for many academic subjects. This support has eased some of the difficulties experienced by students who find concentrating on the content of a lecture while writing at the same time a barrier to learning. Research has shown that actively reviewing notes that are not necessarily made by students, can still result in successful outcomes. The proviso being that the interactions with lecture content and knowledge that needs to be learnt can happen in a way that enhances long term memory.Design/methodology/approachLecturers in higher education often use a variety of e-learning systems to offer varying forms of gamification to support remembering, understanding and application of content taught during lectures. However, these systems may not necessarily enable the practice of higher order learning skills as will be demonstrated with a review of the features offered by 37 online platforms used in colleges and universities.FindingsIt is proposed that with the support of generative AI, it is possible to introduce more accessible interactive activities to enable students, including those with disabilities, the ability to analyze, evaluate and create personalized content linked to transcribed lecture notes.Originality/valueThe suggestions offered include the use of online multiformat activities that can be automatically checked for accuracy against the originally accessed transcriptions and their summaries.
Purpose This opinion paper argues that explainable artificial intelligence (XAI) represents a critical bridge for achieving educational equity in higher education for low-resource language communities, but only when implemented through genuine community leadership and cultural sensitivity. Design/methodology/approach Drawing on recent research (2022–2025) and successful indigenous-led AI initiatives, this paper examines the intersection of technical capability, trust dynamics and educational outcomes. It presents an integrated framework combining the CARE Principles for Indigenous Data Governance with the XAI-ED framework for ethical XAI implementation in education. Findings While technical advances have achieved remarkable accuracy (86% for indigenous language recognition), adoption remains limited due to historical mistrust and digital colonialism. Success stories from Te Hiku Media and similar initiatives demonstrate that community-led approaches with transparent AI systems significantly improve educational outcomes while building essential trust. Originality/value This paper contributes a critical perspective on why explainability alone is insufficient; rather, XAI must be embedded within frameworks of indigenous data sovereignty and community empowerment to achieve genuine educational equity.
Purpose - This paper aims to investigate how different forms of artificial intelligence (AI) technologies can be used to support smallholder farmers in climate change adaptation efforts while also benefiting the locavore movement and food sovereignty initiatives. Design/methodology/approach - A technical review of currently existing and adopted AI solutions to determine the performance, adoption framework and conformity to locavores in various geographical regions. The application examples from Asia, Africa and the Middle East included in the study vary in terms of technological application and integration model. Findings - The review revealed knowledge gaps in AI applications for climate adaptation and locavore food systems from which new categories of applications with proven usefulness can be derived. Climate knowledge systems incorporating indigenous knowledge, resource management technologies that sustain indigenous crops, market linkage technologies eradicating exploitative middlemen and biodiversity-preserving technologies that support agricultural diversity. These technologies are friendly to locavore principles as long as they are applied through culturally relevant and locally appropriate methods. Research limitations/implications - The frequent developments in the agricultural use of AI could result in including only some recent implementations in the study. Certainly, there is probably publication bias towards successful implementations, and there remains very little information on any technology due to its relatively recent implementation. Practical implications - Appropriate measures should be taken to adopt AI in agriculture, which includes a bottom-up approach to its deployment, tackling issues of the digital divide, data ownership and sovereignty and respecting the indigenous knowledge system. Technologies have been developed hand in hand with farmers so that interfaces used in those applications reflect the illiterate levels and are also useable in areas where the Internet is rare. Originality/value - The subject of the current paper is the areas of AI technologies, climate adaptation and the locavore movements, with a focus on small-scale farmers. Thus, the paper focuses on innovation and enactment issues that can help explain how AI can enhance rather than eliminate indigenous farming practices.
PurposeThis paper aims to provide a systematic research overview of the effects of artificial intelligence (AI) on the educational performance of children with neurodevelopmental disabilities (NDD) following the preferred reporting items for systematic reviews and meta-analyses statement.Design/methodology/approachThe selected studies, published between November 2021 and November 2024, were identified through the search of SCOPUS, EBSCO, PsycInfo, ProQuest and two preprint-servers related to education. The selection criteria were based on (1) articles focusing on children and youth (ages 5-17 years), (2) articles focusing on children with NDD, (3) articles addressing student achievement, (4) articles reporting on studies that collected primary data and (5) articles reporting on studies that used school/researcher-administered assessments (objective) or self/hetero-reported measures (subjective). The screening of titles, abstracts and keywords left a final sample of n = 15 scientific papers.FindingsThe studies highlight outcomes classified into four primary categories: cognitive processing, academic performance, engagement and motivation and social and emotional learning. The most important findings are concerned with the difficulties faced by children in attention, memory, logical reasoning, reading, tasks related to science, technology, engineering, mathematics and in some cases engagement and motivation.Practical implicationsEmpirically, the study outlines pathways for future research on AI in special education (SE). In practice, the study presents challenges in implementing AI in SE and discusses practical and policy implications.Originality/valueTo summarize, studies (e.g. Drigas and Ioannidou, 2013; Hopcan et al., 2023; Yang et al., 2024) provide valuable insights into AI's transformative potential in SE, however, most studies are limited to specific disabilities, technologies or demographic contexts, often neglecting a broader synthesis of AI's impact on NDD.
Purpose The purpose of this paper is to review studies relevant to the early identification of behavioural changes in children with autism when they start to feel overwhelmed and use their detection for the timely launch of evidence-based well-being interventions for their mitigation. Design/methodology/approach A scoping review of electronic databases was undertaken focusing on the studies related to emotions or arousal detection in children with Autism. Findings The literature selected for this research explores different methodologies used for detecting precursors of emotional dysregulation (termed meltdown) in individuals with autism. Results suggest that use of multi-modal data is an under researched area, further compounded by the lack of accessible state of the art data sets. Originality/value This paper is first of its type aiming to review and categorise existing literature related to approaches for emotion and arousal detection in children with Autism to identify gaps where contemporary computer science methods may help.
