Research On The Methods And Application Of Handwriting Movement Analysis In Children

For the rich cognitive,physiological and pathological information contained in handwriting movement,it has been widely analyzed to get an insight of the characteristics of human cognitive and fine motor control,as well as the underlying mechanism of related diseases.Although much effort has been made on the research of handwriting movement analysis,previous study mainly focus on the disease population.Few attention has been paid on healthy group,and the description of handwriting feature is limited.Moreover,application level research of handwriting movement analysis is insufficient at present.Hence,we introduced a new handwriting consistency aspect and analyzed writing features of healthy children,as well as autism spectrum disorders(ASD)children respectively.Meantime,the machine learning algorithm was introduced for realizing the application in quantitative classification of ASD.The main research contributions are as follows:1.Chinese character writing analysis of healthy childrenFor the quantitative assessment of handwriting difficulty,it is of great importance to understand the characteristics of handwriting development in healthy children,.At present,only a few attention has been paid on healthy children.And the study on handwriting development based on Chinese character writing is scare.Thus,we used Chinese character writing task to analysis the developmental feature of handwriting for Chinese children from grade 1 to 5,and to analysis the influence of age and gender factors.Based on repetitive character writing,the standard deviation of spatial-temporal,kinematic and dynamic measures were calculated for the evaluation of handwriting consistency.Moreover,the Dynamic time warping(DTW)algorithm was introduced to examine the consistency both in the writing trajectory and in the writing velocity profiles.The results showed that the development of handwriting skill was nonlinear with the grade and age increasing.With the increase of grade,writing time and magnitude decreased gradually,and the writing pressure and handwriting consistency improved.The findings of consistency features suggest that handwriting consistency is a crucial aspect in the evaluation of children’s handwriting ability and the proposed consistency measures including DTW distances are new effective indicators in quantify handwriting ability of developmental children.2.Analysis of handwriting performance in children with ASDThough children with ASD face a high risk of handwriting difficulty,research on handwriting feature of ASD is limited.Besides,although the developmental characteristic of handwriting consistency has been known,the influence of related disease on handwriting consistency is still not clear.In this study,we designed Arabic numeral handwriting tasks and examined the handwriting performance of Chinese ASD children for the first time.The spatial-temporal,kinematic and dynamic measures were calculated for ASD and typical developing(TD)groups.For repetitive writing tasks,consistency measures including DTW distance were extracted.Compared with the TD group,the heterogeneity handwriting feature for ASD children was reflected by larger size,higher velocity and higher acceleration in the numeral sequence task,and less consistency in repetitive writing task.The findings indicate that abnormal handwriting performance is likely associated with the complexity of the writing task.The results further verified the effectiveness of consistency measures,including DTW distances,in handwriting evaluation.Besides,it offers a reference for target teaching and intervention of ASD.3.Quantitative classification research using handwriting featuresAssisted diagnosis is a main concern in the applications of handwriting movement analysis.Nevertheless,there is limited research on hand-written motion analysis of practical application problems.Based on the handwriting characteristics of ASD children in the Arabic numeral sequence writing task,we applied machine learning algorithm for group classification to realize the application of handwriting in quantitative classification of disease population.To improve the classification accuracy,image features,which are able to capture the local character of handwriting,were proposed.Then,combining with traditional the spatial-temporal and kinematics characteristics,we developed a new multiple distance metric learning method to effectively learn different matrixes for the obtained features,which leads to the final classification.The influence of different features on accuracy was analyzed by experiment.Meanwhile,the classification performance was compared between the classical classification algorithms and the proposed method.The experimental results suggest that the combination of traditional features and image features using the proposed method can achieve a better classification.Besides,this proposed algorithm provides a new approach for clinical auxiliary diagnosis and comprehensive evaluation of ASD and other related diseases.