EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Developing a Large Language Model-Based Feedback Tool for Primary School Teachers Supporting Neurodivergent Learners

Reno Yuri Camilleri, Vanessa Camilleri, Matthew Montebello

AI, Computer Science and Robotics Technology · 26.03.2026 · 5

DOI: 10.5772/acrt.20260007 ↗

Kurz gefasst

Fachartikel in der Zeitschrift AI, Computer Science and Robotics Technology, online erschienen am 26.03.2026. Verfasst von Reno Yuri Camilleri und 2 weiteren Autor:innen. Die Arbeit nennt 16 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Primary school teachers frequently report challenges when supporting pupils with attention deficit hyperactivity disorder (ADHD) and autism spectrum disorder (ASD), yet access to realistic, practice-based professional development (PD) remains limited. This study presents early results from the design and evaluation of a locally run, data-driven Large Language Model (LLM) intended to simulate classroom interactions and provide formative feedback to teachers. The project aimed to (1) synthesise best-practice evidence for effective responses to neurodivergent pupils, (2) examine reliable methods for prompt engineering and deploying LLMs in controlled educational environments, and (3) develop a scenario-driven tool capable of analysing teacher actions during simulated exchanges. The methodology combined a systematic review of intervention literature with the design-based development of a prototype system. Ten test scenarios were generated for pupils aged 6–10 (five ADHD and five ASD), with scenarios and feedback grounded in curated academic sources. The model was hosted entirely offline for privacy and ethical safeguards, and prompts were structured to generate concise scenarios, maintain consistent pupil behaviour, and produce research-aligned feedback. Early qualitative findings indicate that the system reliably applied best-practice principles when evaluating teacher responses. These preliminary results suggest that an LLM-based training tool can meaningfully support reflective practice by offering repeatable, evidence-informed simulations. With further refinement, broader testing, and increased scenario diversity, the tool has the potential to enhance PD opportunities and improve teaching and learning outcomes for neurodivergent pupils in primary schools.

Zitieren

APA

Camilleri, R. Y., Camilleri, V., & Montebello, M. (2026). Developing a Large Language Model-Based Feedback Tool for Primary School Teachers Supporting Neurodivergent Learners. AI, Computer Science and Robotics Technology, 5. https://doi.org/10.5772/acrt.20260007

Vancouver

Camilleri RY, Camilleri V, Montebello M. Developing a Large Language Model-Based Feedback Tool for Primary School Teachers Supporting Neurodivergent Learners. AI, Computer Science and Robotics Technology. 2026;5. doi:10.5772/acrt.20260007

BibTeX
@article{Camilleri2026,
  author = {Camilleri, Reno Yuri and Camilleri, Vanessa and Montebello, Matthew},
  title = {Developing a Large Language Model-Based Feedback Tool for Primary School Teachers Supporting Neurodivergent Learners},
  journal = {AI, Computer Science and Robotics Technology},
  year = {2026},
  volume = {5},
  doi = {10.5772/acrt.20260007}
}
Kostenlos starten →