A Bayesian computational model for online character recognition and disability assessment during cursive eye writing

Area of Science:

• Neuroscience
• Human-Computer Interaction
• Biomedical Engineering

Background:

• Traditional writing methods are inaccessible to individuals with severe motor impairments.
• Eye-tracking technology offers a potential alternative for communication and interaction.
• Existing models for analyzing human movement lack specificity for eye-based writing.

Purpose of the Study:

• To introduce and validate a novel apparatus for "eye writing" using illusory visual stimuli.
• To develop and apply a probabilistic model (Bayesian Action-Perception for Eye On-Line model - BAP-EOL) for analyzing eye-writing trajectories.
• To explore the potential of eye writing for motor-impaired patients, including character recognition and disability assessment.

Main Methods:

• Development of a novel apparatus presenting illusion-inducing visual stimuli to guide eye movements.
• Adaptation and application of a probabilistic model (BAP-EOL) encoding letter trajectories, size, high-frequency components, and pupil diameter.
• Utilizing Bayesian inference for tasks including letter recognition, novelty detection, and disability assessment.

Main Results:

• Demonstrated feasibility of character recognition using the eye-writing apparatus and BAP-EOL model.
• Successfully implemented novelty detection for identifying unknown symbols within eye-written trajectories.
• Showcased the capability of disability assessment by analyzing fine motor control characteristics during eye writing.

Conclusions:

• Eye writing presents a viable communication and assessment tool for motor-impaired individuals.
• The BAP-EOL model effectively processes eye-writing data for recognition, detection, and assessment.
• Further research is needed to address technical challenges and optimize eye-writing for clinical applications.