News Story

Explainable AI study on Education Equity awarded Best Paper at ICAITE 2025

Thursday 5 Feb 2026

A paper titled “Explainable AI Approach to Compare Predictive Drivers of Student Success for Disabled and Non-Disabled Learners”, authored by Dr Retno Larasati from the Shifting Power Project in KMi, received the Best Paper Award at the International Conference on Artificial Intelligence and Teacher Education (ICAITE 2025). The conference, an international gathering on AI and education, was held from 7–9 December 2025 in Kyoto, Japan.

The research focuses on understanding and addressing the needs of diverse student populations, including those with disabilities, in both traditional and online learning. The study employs Explainable AI (XAI) to identify the factors that influence success for different groups of learners. To explore the reasons behind varying student outcomes, the researchers used an XAI framework to analyse predictive models and make their decision processes transparent and understandable.

Drawing on the Open University Learning Analytics Dataset (OULAD), a previous KMi project, the paper applied the XAI method SHAP, which assigns an importance value to each feature for every prediction. This enables researchers to understand the contribution of individual factors to a model’s output. The study carried out a comparative analysis between students identified as having disabilities and those without. The findings reveal divergences in the predictive drivers of academic success between these two groups. For example, factors that predict success for a non-disabled learner may have a different impact on a student with disabilities.

The implications of these findings are significant. The paper demonstrates that XAI is essential for designing equitable educational support systems. By understanding the drivers of success for students with disabilities, educators can tailor strategies to meet individual needs. This may include personalised learning interventions, adaptive course materials, or specific teaching approaches that are more effective for disabled learners.

Recognition at ICAITE 2025 highlights the timely nature of this research. As educational institutions expand online learning and pursue greater inclusivity, insights from this Explainable AI approach will be invaluable. It supports educators in creating supportive, effective, and equitable learning environments that help every student succeed.

This achievement represents an important contribution to Artificial Intelligence, Learning Analytics, and Teacher Education, supporting the development of educational technology and policy focused on student success and equity.