Abstract:

The inclusive education paradigm requires ensuring equitable access to the curriculum and fair assessment for all students. Students with Specific Learning Disabilities (SLDs), such as dyslexia, dysorthographia, dyscalculia, and attention deficit hyperactivity disorder, as well as students with autism spectrum disorder and mild intellectual disability, require precise and context-specific assessment tools. The current process used to define these measures is often insufficiently standardised, relies heavily on individual teacher expertise, and demands substantial time to ensure alignment with national educational guidelines.

This article summarizes the development and validation of the Intelligent Pedagogical Adaptation Agent (IPAA), a specialized artificial intelligence tool that supports educators and multidisciplinary teams. The IPAA analyzes pedagogical variables and assessment materials, integrates the student’s learning profile and the Portuguese legal framework, and automatically generates individualized and legally compliant accommodations.

The IPAA is grounded in the principles of the Decree-law n. º 54/2018 (Legal Framework for Inclusive Education), which serves as the central regulatory authority. The agent also integrates best scientific practices from neuropsychology, special education research, and Universal Design for Learning (UDL), ensuring evidence-based and maximally effective measures.

The Agent employs a structured, interactive protocol:
- Guided Data Elicitation: Systematically prompts the user for necessary data, including pedagogical context, student profile (clinical/academic/history), logistical resources, and motivational elements.
- Assessment Analysis: User submits the test material. The IPAA analyzes question complexity (e.g., reading load) against the student's identified difficulties (e.g., phonological deficits, working memory constraints).
- Accommodation Generation: The system generates adaptation proposals tailored to the student’s learning profile and to the cognitive demands of each assessment component. The model consistently produces both global recommendations and item-level analyses by default, while still allowing the user to interact with the system to optimise or refine the outputs when needed. Although the tool aims to minimise user prompting, some iterative clarification may occasionally be required to produce the final accommodation report.
- Automated Reporting & Justification: Final output is a professional Accommodation Report. It lists recommended measures (e.g., time, assistive technology) and provides a detailed rationale, citing the Decree-Law n. º 54/2018 and scientific articles, significantly easing school bureaucracy.

Goals:
The IPAA is designed to generate highly personalized accommodations that align with each student's unique profile, to improve the assessment of students with special educational needs by ensuring that scores reflect knowledge rather than disability, to guarantee compliance with DL 54/2018 through transparent and auditable recommendations, and to streamline pedagogical practice by reducing the planning burden on educators.
The IPAA constitutes an important advance in the integration of artificial intelligence into special education, offering a structured model for promoting equity, legal compliance, and quality. The next stage involves an experimental pilot study to evaluate its effectiveness on student performance and user acceptance to refine the IPAA.

Keywords:

Inclusive Education, Artificial Intelligence (AI), technology, pedagogic innovation.