Handwriting difficulty screening tool based on dynamic data from drawing process

Abstract

Children with handwriting difficulty are advised to join an intervention program to rectify the problem at an early stage. However, the available screening tools suffer from subjectivity judgement while lack of expertise reduces the chance for every student to be screened. Yet, digitalized screening tools that use dynamic data from writing activities are only applicable to those who know the language. These limitations had led this study to develop an objective handwriting difficulty screening tool based on dynamic data of drawings. Three attributes extracted from 120 sets of dynamic data from drawing process were found to be significant in differentiating below-average writers from average writers. The attributes were then used to train Support Vector Machine prediction model. To test the validity and reliability of the prediction model, additional sets of data were acquired from 36 pupils. The performance of the tool was compared with the results from the Handwriting Proficiency Screening Questionnaire (HPSQ) that employs teachers’ observations on pupils’ handwriting ability. With 78% reliability, 69% of the predictions made by the developed tool was in accordance with the teachers’ observation. Most importantly, 53% of the average writers were screened as having handwriting problems. This denotes the objectivity of the developed tool in identifying below-average writers who failed to be recognized through teacher’s observation.