Related Experiment Video
Updated: Jun 13, 2026
Using Eye-tracking to Assess the Relative Importance of Visual and Vestibular Input to Subcortical Motion Processing in the Roll Plane
Published on: August 22, 2025
Enhancing the detection of visual-cognitive disruptions in automated dyslexia diagnosis using viscoelastic
S Regilan1, R Indumathi2, J Jenitha3
1Associate Professor, Department of Electronics and Communication Engineering, Asan Memorial College of Engineering and Technology, Chengalpattu,Chennai, Tamil Nadu, India. lssregilan@gmail.com.
Purpose:
Dyslexia can be assessed using eye-tracking technology to monitor reading patterns, but challenges persist in distinguishing dyslexia-specific anomalies from those caused by other factors due to individual reading variability.
Methods:
To address these challenges, enhancing the Detection of Visual-Cognitive Disruptions in Automated Dyslexia Diagnosis Using Viscoelastic Constitutive Artificial Neural Networks for Eye Movement Analysis (DVDD-VCANN-EMA) is proposed. Firstly, input data is gathered from ETDD70: Eye-Tracking Dyslexia Dataset. Then the input data is pre-processed employing Continuous-Discrete Derivative-Free Extended Kalman Filter (CDDFEKF) which is utilized to clean the data and normalization. Then the pre-processed data are given to High- Order Time-Reassigned Synchrosqueezing Transform (HTSST) for feature extraction. HTSST is employed to extract relevant features such as word vectors, eye movement metrics, and saliency maps. The extracted features are input into the Viscoelastic Constitutive Artificial Neural Networks (VCANN),AQ2 classify such as dyslexic and non-dyslexic. In general, VCANN do not demonstrate the use of optimization strategies to determine weight parameters for accurately diagnosing dyslexia through eye movement analysis. Therefore, the Doll maker Optimization Algorithm (DOA) is employed in this study to optimize the weight parameters of VCANN.
Results:
The proposed technique implemented in python, demonstrates substantial improvements in accuracy, precision, recall, F1 score, Specificity and AUC. The DVDD-VCANN-EMA model achieves peak performance with 98.5% accuracy, 98.5% recall, and 98.5% F1-score, and the fastest computation time of 1.170 s.
Conclusion:
for effective compare with existing methods such as Dyslexia Analysis and Diagnosis according to Eye Movement (DAEM-CNN), INSIGHT: Combining Fixation Visualisations and Residual Neural Networks for Dyslexia Classification From Eye-Tracking Data (FVDEDNN) and Optimal Ensemble Learning Model for Dyslexia Prediction Based on an Adaptive Genetic Algorithm (OEDP-SVM).
