Page 646 - AI for Good Innovate for Impact

AI for Good Innovate for Impact

Use case- 6: AI-Powered Personalized Learning for Neurodivergent

Students

Country: Ethiopia

Organization: Addis Ababa Science and Technology University
Contact Person(s): Firaol Teshale Negera (firaolteshale@ gmail .com)

1 Use Case Summary Table

Category Education
The problem to be Traditional classrooms fail to adapt to neurodivergent learners’ cognitive
addressed and emotional needs, leading to disengagement, anxiety, and poor learning
outcomes.

Key aspects of the Education (Education Technology, Accessibility, Mental Health)
solution Real-time adaptive learning assistant using AI to detect engagement,
stress levels, and customize content formats. Combines emotion detection,
NLP-based content generation, and accessibility tools.
Technology NLP, Transfer Learning, RAG (Retrieval-Augmented Generation), Facial
keywords Expression Analysis, Sentiment Analysis, Adaptive Interfaces, Multimodal AI.
Data availability Public datasets (Affect Net, SLDR, Kaggle Learning Data); private data
collected with consent during pilots.

Metadata (type of Eye tracking, facial expression data, learning interaction logs, emotion-la-
data) belled feedback.

Model training Fine-tuning pre-trained models for emotion recognition and adaptive
and fine-tuning content delivery; RAG for context-aware content personalization

Testbeds or pilot Planned pilot testing in special education classrooms with validation by
deployments educational psychologists and therapists

Code Reposito- Will be available upon completion of prototype on GitHub
ries

2 Use Case Description

2�1 Description

AI-Powered Personalized Learning for Neurodivergent Students

Millions of neurodivergent students including those with autism, ADHD, dyslexia, and sensory
processing disorders face learning barriers in traditional classroom settings. Standardized

610