An Efficient Machine Learning-Based Feature Optimization Model for the Detection of Dyslexia.

Area of Science:

• Neuroscience
• Computational Psychology
• Developmental Psychology

Background:

• Dyslexia is a prevalent childhood neurological disorder.
• Early and accurate detection of dyslexia is crucial for effective intervention.
• Existing detection methods can be complex and time-consuming.

Purpose of the Study:

• To develop and evaluate a machine learning model for dyslexia detection.
• To utilize data from a unified gaming test for improved diagnostic accuracy.
• To explore the efficacy of Support Vector Machines (SVM) and Artificial Neural Networks (ANN) for dyslexia identification.

Main Methods:

• A unified gaming test was administered to children with and without dyslexia.
• Principal Component Analysis (PCA) was used to reduce data dimensionality.
• Support Vector Machine (SVM) with various kernels and Artificial Neural Network (ANN) models were trained and tested.
• Model performance was evaluated based on detection accuracy.

Main Results:

• SVM with Radial Basis Function (RBF) kernel achieved up to 93% accuracy with 3 principal components.
• Artificial Neural Network (ANN) demonstrated superior performance, reaching 95% accuracy with 3 principal components.
• The proposed machine learning approaches showed high efficacy compared to existing methods.

Conclusions:

• Machine learning models, particularly ANN, can effectively detect dyslexia using gaming test data.
• Dimensionality reduction via PCA enhances model efficiency and performance.
• This gaming-based approach offers a promising, accurate, and potentially less complex method for dyslexia detection.