AR numeracy app designed for children with learning disabilities

Researchers at Vellore Institute of Technology in India have developed a web-based augmented reality application designed to help children with learning disabilities learn basic numeracy skills. The study, published in Frontiers in Education, utilized the ADDIE instructional design model grounded in Mayer’s Cognitive Theory of Multimedia Learning to create a tool that specifically targets the visual-symbolic processing weaknesses identified in the target population. This pilot study involved 70 children from four special schools in Chennai, Tamil Nadu, and focused on establishing a traceable pathway from empirical assessment to interface design.

The primary objective of the research was to address the gap in educational technology where augmented reality tools for children with learning disabilities are rarely designed based on an empirical assessment of the learners’ specific numeracy profiles. The team aimed to determine the performance levels of these children across different analytical domains and to explore how the ADDIE model, with CTML applied at the design stage, could guide the development of an effective AR numeracy application. The study also sought to understand how children rated their attitudes toward mathematics and their reception of the AR activity after completing the interaction.

The study employed a single-group exploratory pilot design, which is intended to explore feasibility and acceptability rather than to establish instructional effectiveness through comparative outcomes. A total of 70 participants, with a mean age of 9.17 years, were recruited from four schools for children with special educational needs. The sample consisted of 51 boys and 19 girls, and the selection of schools was purposive within convenience. All participants were treated in accordance with the code of ethics, and the study was approved by the Institutional Ethics Committee for Human Studies at Vellore Institute of Technology.

To establish a baseline for the design process, the researchers administered a ten-item, paper-based numeracy assessment to all participants before any exposure to the augmented reality prototype. This assessment covered five skill areas: visual discrimination, symbol recognition, shape recognition, comparative numerosity, and counting. For statistical analysis, these skills were grouped into three analytical domains: visual-symbol-shape, comparison, and counting. The assessment was administered without a time limit to minimize test-related anxiety, and a teacher was present to read items aloud if necessary, though they were prohibited from assisting with the actual calculations or counting.

The results of the baseline assessment revealed that performance was uneven across the three analytical domains. The mean accuracy for the visual-symbol-shape domain was 72.9%, which was significantly lower than the accuracy for the comparison domain at 82.9% and the counting domain at 83.2%. Statistical analysis using the Friedman test and Bonferroni-adjusted Wilcoxon signed-rank tests confirmed that visual-symbolic processing was the relatively weaker domain in this sample. This specific weakness was then translated into the design of the AR interface, with features specifically targeting visual-symbolic processing.

The development of the AR prototype followed the five stages of the ADDIE model: analysis, design, development, implementation, and evaluation. In the design phase, the interface was grounded in Mayer’s Cognitive Theory of Multimedia Learning, which posits that meaningful learning occurs when learners actively process information through distinct verbal and visual channels. The design incorporated principles such as coherence, signaling, and spatial contiguity to reduce extraneous cognitive load and manage essential processing. For instance, the interface used a minimalistic layout with red dots and arrows to guide attention toward tracing tasks, and it provided synchronized audio-visual feedback to support the learning of number formation.

The resulting prototype was a WebAR platform built using the MyWebAR no-code, browser-based tool, which allowed the application to be accessed directly through a mobile device’s web browser without requiring a dedicated app download. This delivery method was chosen for its accessibility and low technical barrier in a school setting. During the implementation phase, each child interacted individually with the AR activity on a mobile device or tablet, completing a dot-joining task for numbers from 1 to 9. The activity included a marker-based tracking approach and prerecorded human voice pronunciation to reinforce number recognition and formation.

Following the AR interaction, the children completed an eight-item perception survey using a five-point Likert scale. The survey measured their attitudes toward mathematics, their engagement with the activity, and their specific responses to the AR experience. Principal components analysis of the survey data identified two distinct dimensions: general attitude toward mathematics and response to the AR activity. Both dimensions were rated significantly above the scale midpoint, with the mathematics attitude subscale having a mean of 4.09 and the AR activity subscale having a mean of 4.30. These findings suggest that the multisensory, visual-symbol-focused design was well received by the target group.

The researchers noted that the study’s contribution lies in the traceable pathway from an empirically assessed numeracy weakness to a specific set of CTML-derived interface features. Rather than simply applying the ADDIE model, the study demonstrated how a baseline assessment could inform the design rationale for an AR tool. The findings indicate that an assessment-driven, theory-guided design pathway is practically achievable and can be well received by children with learning disabilities in a school-based setting. The prototype established preliminary feasibility and acceptability for this specific type of educational technology.

However, the study has several limitations that must be considered. The design was cross-sectional and single-session, with no control group or pre-post comparative test to establish the effectiveness of the AR intervention in improving numeracy skills. The sample size, while reasonable for a pilot study, showed a proportional imbalance in gender, which may limit the generalizability of the findings. Additionally, the internal consistency of the numeracy assessment was modest, and ceiling effects were substantial, with 44% to 60% of children achieving maximum scores in each domain. Therefore, the domain scores are best interpreted as group-level indicators of relative difficulty rather than as reliable measures of individual ability.

The authors suggest that future research should build on these findings through longitudinal and multimethod approaches. Controlled, longitudinal studies could help validate the numeracy assessments and compare learning outcomes to determine the true effectiveness of the AR design. Future versions of the tool should also expand beyond single-digit numbers to include larger digit ranges and a broader range of arithmetic operations. Furthermore, the system should be modified to securely record individual performance data, such as scores and time-on-task, to allow for a more detailed analysis of progress over time.

This research highlights the potential of augmented reality to support children who struggle with conventional teaching methods by providing a personalized and engaging learning environment. By combining instructional design models with cognitive learning theories, the study offers a reusable design pathway for developing numeracy tools for this population. While the findings are preliminary, they provide initial evidence that AR-based activities can be both feasible and acceptable for children with learning disabilities in a mathematics learning context.

Subject of Research: Right way to write numbers: designing an augmented reality-based numeracy application for children with learning disabilities using the ADDIE model

Article Title: Right way to write numbers: designing an augmented reality-based numeracy application for children with learning disabilities using the ADDIE model

Article References: Jana, T., & Saravanan, S. (2026). Right way to write numbers: designing an augmented reality-based numeracy application for children with learning disabilities using the ADDIE model. Frontiers in Education, 11, Article 1961050. https://doi.org/10.3389/feduc.2026.1961050

Image Credits: AI Generated

DOI: 10.3389/feduc.2026.1961050

Keywords: Right, write, numbers, designing, augmented, reality-based, numeracy, application, children, learning, disabilities, ADDIE