Date of Submission

Fall 12-10-2023

Degree Type


Degree Name

Master of Science in Computer Science (MSCS)


Computer Science

Committee Chair/First Advisor

Dr. Selena He



Neuroscience and Psychology


Dr. Selena He

Committee Member

Dr. Joy Li

Committee Member

Dr. Sungchul Jung


In the realm of educational technology, attention training plays a critical role in tailoring learning experiences to individual needs, especially for learners with autism spectrum disorder (ASD). This study presents an advanced framework that provides insights into the efficacy of reinforcement strategies by predicting their impact on attention enhancement in educational virtual reality (VR) settings, utilizing physiological biomarkers such as eye-tracking (ET), heart rate (HR), and electrodermal activity (EDA). A comprehensive comparative analysis was undertaken to evaluate the performance metrics of various machine learning (ML) and deep learning (DL) algorithms. The results showcased the robustness of gradient boosting (GB) and random forest (RF) in predicting the impact of reinforcement training in attention increase with high F1-score and ROC\_AUC values. GB achieved remarkable performance on all features dataset with 77.7\% F1-score and 77.08\% ROC\_AUC, while RF excelled on selected features dataset with 80\% F1-score and 81.94\% ROC\_AUC. The study also explores pattern recognition between autistic and non-autistic individuals, providing insights into the distinctive attentional profiles. An LSTM time-series model was also developed for real-time prediction, offering a pathway for personalized and adaptive learning experiences. The integration of artificial intelligence (AI) models and physiological data holds significant promise for enhancing attention training, with implications extending to personalized education for ASD. The study sets the stage for future enhancements in LSTM prediction accuracy and the development of real-time, tailored educational interventions.

Available for download on Friday, January 10, 2025