Automated human-level diagnosis of dysgraphia using a consumer tablet

Abstract

The academic and behavioral progress of children is associated with the timely development of reading and writing skills. Dysgraphia, characterized as a handwriting learning disability, is usually associated with dyslexia, developmental coordination disorder (dyspraxia), or attention deficit disorder, which are all neuro-developmental disorders. Dysgraphia can seriously impair children in their everyday life and require therapeutic care. Early detection of handwriting difficulties is, therefore, of great importance in pediatrics. Since the beginning of the 20th century, numerous handwriting scales have been developed to assess the quality of handwriting. However, these tests usually involve an expert investigating visually sentences written by a subject on paper, and, therefore, they are subjective, expensive, and scale poorly. Moreover, they ignore potentially important characteristics of motor control such as writing dynamics, pen pressure, or pen tilt. However, with the increasing availability of digital tablets, features to measure these ignored characteristics are now potentially available at scale and very low cost. In this work, we developed a diagnostic tool requiring only a commodity tablet. To this end, we modeled data of 298 children, including 56 with dysgraphia. Children performed the BHK test on a digital tablet covered with a sheet of paper. We extracted 53 handwriting features describing various aspects of handwriting, and used the Random Forest classifier to diagnose dysgraphia. Our method achieved 96.6% sensibility and 99.2% specificity. Given the intra-rater and inter-rater levels of agreement in the BHK test, our technique has comparable accuracy for experts and can be deployed directly as a diagnostics tool.

Keywords: Diagnosis; Patient education.

Fig. 1

Box plot representing the F1-score as a function of time period of the test used for training. We used Random Forest for classification, and each model was trained following the same k-fold, cross-validation procedure with k = 25

Fig. 2

Distribution of the dysgraphic children (D dataset) and the non-dysgraphic children (TD dataset). For static features: Bandwidth of Tremors Frequencies and the Space Between Words features. For kinematic features: Median of Power Spectral of Speed Frequencies and the In Air Time Ratio features. For tilt features: Median of Power Spectral of Speed of Tilt-y change and the Bandwidth of Power Spectral of Speed of Tilt-x change features

Fig. 3

A comparison of different metrics for a non-dysgraphic child (left) and a child with dysgraphia (right)

Fig. 4

The whole process used to extract the frequency spectrum of our signal. (1) We first divided the BHK text into bins of 600 points. (2) For each packet, the signal was extracted. (3) We then computed the Fourier transform of the signal. (4) We took the average of all signals and finally performed a normalization. At the top of the figure is presented an example of a signal extracted from the data: the red dots are the point coordinates recorded by the device during handwriting. The vectors in blue are “local” vectors linking two consecutive points. The vector in green is the “global” vector (average of the nine blue vectors) representing the global direction of the handwriting. The cross product of these two vectors gives us an indication of the smoothness/shakiness of the handwriting. The image on the right comes from a writer with smoother/less shaky handwriting than the one on the left. The cross product operation will detect this difference