Autism spectrum disorder-level prediction and personalized education planning using TabNet

Area of Science:

• Neuroscience
• Educational Psychology
• Computer Science

Background:

• Autism spectrum disorder (ASD) significantly impacts students' social, communication, and behavioral skills.
• Current research often focuses on diagnosis, lacking effective methods for predicting ASD levels and tailoring educational strategies.
• Personalized education plans are crucial for addressing diverse support needs in autistic students.

Purpose of the Study:

• To develop an autism-level categorization system for students with autism spectrum disorder.
• To create a tailored education plan recommendation system linked to predicted autism levels.
• To integrate autism-level prediction with personalized educational strategy development.

Main Methods:

• Collected a dataset including verbal ability, social interaction, sensory sensitivity, and attention span for students with ASD.
• Preprocessed features and trained a TabNet model for autism level categorization.
• Developed a web application to recommend personalized education plans based on model predictions.

Main Results:

• Achieved 99.37% accuracy and 98.91% precision in autism level categorization.
• Successfully integrated autism level prediction with personalized education plan recommendations.
• Demonstrated the model's effectiveness in categorizing autism levels and suggesting tailored educational strategies.

Conclusions:

• The proposed system effectively categorizes autism spectrum disorder levels.
• The integrated approach provides personalized education plan recommendations for autistic students.
• This research enhances educational effectiveness and personalization for students with autism spectrum disorder.