Dysgraphia disorder forecasting and classification technique using intelligent deep learning approaches.

Area of Science:

• Neuroscience
• Developmental Psychology
• Computer Science

Background:

• Dysgraphia is a learning disability impacting writing abilities, often challenging to diagnose early in childhood.
• Early identification of dysgraphia is crucial for timely intervention and support.
• Existing methods for dysgraphia detection may lack sufficient accuracy or rely on complex assessments.

Purpose of the Study:

• To develop and evaluate a novel, highly accurate method for detecting dysgraphia in children using handwriting analysis.
• To leverage deep transfer learning and a specialized mathematical model for feature extraction and classification.
• To compare the proposed method's efficacy against traditional machine learning and deep learning approaches.

Main Methods:

• Handwriting and geometric features were extracted from images using the Kekre-Discrete Cosine Transform (K-DCT).
• Deep transfer learning was employed for effective feature learning and dysgraphia identification.
• The proposed Kekre-Discrete Cosine Transform with Deep Transfer Learning (K-DCT-DTL) approach was implemented and tested.

Main Results:

• The K-DCT-DTL approach demonstrated a high accuracy rate of 99.75% in detecting dysgraphia from handwritten images.
• The proposed method significantly outperformed existing machine learning and deep learning techniques in accuracy.
• Handwritten image analysis proved effective for identifying children with dysgraphia.

Conclusions:

• The K-DCT-DTL method offers a highly efficient and accurate solution for the early detection of dysgraphia.
• This approach shows significant potential for improving diagnostic capabilities in developmental learning disorders.
• Utilizing handwritten drawings provides a viable and effective pathway for identifying dysgraphia in pediatric populations.