Exploring Adaptive Learning Design Through Rshiny: An IRT-Based Approach for MathematicsEducation in Sri Lanka

Mathematics consistently records the lowest pass rates at the G.C.E Ordinary Level (O/L) examinations in Sri Lanka, underscoring the urgent need for innovative approaches to address the wide variation in student ability. Globally, adaptive learning has been advanced through intelligent tutoring systems, Bayesian frameworks, and rule-based models: however, its adoption in Sri Lanka secondary education remains minimal. This study presents the design of an adaptive learning framework developed as a proof of concept rather than a fully operational system implemented using R-shiny. The framework employs Item Response Theory (IRT), specially the three-parameter logistic (3PL) model, which incorporates item difficulty, discrimination, and guessing, thereby offering a robust representation of student performance. Student ability estimates are uploaded recursively after each response, while subsequent items are selected using maximum information criteria to enhance both efficiency and measurement precision. Simulation results demonstrate that the proposed IRT-based design can efficiently distinguish among students at different ability levels and generate reliable estimates of performance. These findings highlight the potential of prototype-level adaptive learning designs to provide a statistically rigorous and scalable foundation for advancing mathematics education in Sri Lanka, while also establishing a pathway toward future system development and classroom integration.