Adaptive severity-aware educational content recommendation for autistic learners using a LinUCB-based contextual bandit framework

Despite the rapid growth of digital learning technologies, many existing educational platforms remain insufficiently adaptive to the diverse behavioural and cognitive characteristics exhibited by autistic learners. To address this challenge, this study proposes a LinUCB-based contextual bandit framework for personalized educational content recommendation using learner-specific contextual representations. The proposed proof-of-concept framework constructs severity-aware and age-aware learner profiles from publicly available Autism Spectrum Disorder (ASD) screening datasets obtained from the UCI Machine Learning Repository and Kaggle and complements them with controlled synthetic educational interaction variables to investigate adaptive recommendation strategies. The resulting contextual representation integrates demographic information, behavioural indicators, severity-related characteristics, and educational interaction attributes to support recommendation across multiple learning modalities. Experimental evaluation using stratified 10-fold cross-validation demonstrates that the proposed framework consistently outperforms conventional supervised learning algorithms and alternative bandit-based approaches, achieving a recommendation accuracy of 87.3%. Comparative benchmarking and feature ablation analyses further highlight the contribution of contextual learner attributes to recommendation performance and model interpretability. Owing to its computational efficiency and interpretable contextual bandit formulation, the proposed framework provides a promising foundation for future research on explainable, privacy-conscious, and context-aware personalized educational recommendation systems for autistic learners. Future work will focus on validation using authentic learner interaction data together with the incorporation of multimodal behavioural information to further improve recommendation quality and generalizability.

Authors

Institutions

Publication Details

Journal
Scientific Reports
Published
2026-11-18
DOI
https://doi.org/10.1038/s41598-026-74938-x
Primary Topic
Recommender Systems and Techniques
Type
article
Field-Weighted Citation Impact
0.00

Funders

Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
article

Adaptive severity-aware educational content recommendation for autistic learners using a LinUCB-based contextual bandit framework

Sangeeta Sangani, Vijayalaxmi N. Rathod, Rayangouda H. Goudar
3 links
Scientific Reports
Recommender Systems and Techniques
article

Adaptive severity-aware educational content recommendation for autistic learners using a LinUCB-based contextual bandit framework

Sangeeta Sangani, Vijayalaxmi N. Rathod, Rayangouda H. Goudar
article en

Abstract

Despite the rapid growth of digital learning technologies, many existing educational platforms remain insufficiently adaptive to the diverse behavioural and cognitive characteristics exhibited by autistic learners. To address this challenge, this study proposes a LinUCB-based contextual bandit framework for personalized educational content recommendation using learner-specific contextual representations. The proposed proof-of-concept framework constructs severity-aware and age-aware learner profiles from publicly available Autism Spectrum Disorder (ASD) screening datasets obtained from the UCI Machine Learning Repository and Kaggle and complements them with controlled synthetic educational interaction variables to investigate adaptive recommendation strategies. The resulting contextual representation integrates demographic information, behavioural indicators, severity-related characteristics, and educational interaction attributes to support recommendation across multiple learning modalities. Experimental evaluation using stratified 10-fold cross-validation demonstrates that the proposed framework consistently outperforms conventional supervised learning algorithms and alternative bandit-based approaches, achieving a recommendation accuracy of 87.3%. Comparative benchmarking and feature ablation analyses further highlight the contribution of contextual learner attributes to recommendation performance and model interpretability. Owing to its computational efficiency and interpretable contextual bandit formulation, the proposed framework provides a promising foundation for future research on explainable, privacy-conscious, and context-aware personalized educational recommendation systems for autistic learners. Future work will focus on validation using authentic learner interaction data together with the incorporation of multimodal behavioural information to further improve recommendation quality and generalizability.

Scientific Reports
Manipal Academy of Higher Education (IN), National Forensic Sciences University Dharwad Campus (IN), Visvesvaraya Technological University (IN)
Visvesvaraya Technological University
Quality education, Reduced inequalities
Recommender Systems and Techniques
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon