Diagnosis Of Primary School Students' Mathematical Disability Driven By Cognitive Game Data

Mathematical disability is a kind of learning difficulties,which have a great negative impact on student's learning motivation and academic development.The early diagnosis of mathematical disability can help students with math learning difficulties to receive educational intervention as soon as possible,and reduce the negative impact of mathematical disability.The commonly used mathematical disability diagnosis method are still a lack of unified standard of execution and criteria.The cost of manpower and time are too high for conducting a wide range of mathematical disability diagnosis.The traditional cognitive battery used in the diagnosis process is likely to raise the anxiety of primary school students,and cause a lack of motivation,which will have a negative influence to the efficiency of test.This paper is committed to find a rapid and accurate method for large scale mathematical disabilities diagnostic,and verify its effectiveness.To solve the above problem,in the first part of the study,by combing the research results of mathematical disability from home and abroad,it's found that each part of executive function is closely related to mathematical performance,and the deficit of executive function may be the main reason for mathematical disability.Therefore,we put further research on executive function measurement.In the study of cognitive ability measurement in primary school,it is pointed out that adding game elements in the cognitive task can effectively help to keep participants' participation motivation,and reflect the level of their cognitive ability.So we selected some of the common executive function test task and battery,designed game-like cognitive task based on game theory,and developed a set of online cognitive games to reflect the level of executive function using Unity3 D game engine,the reliability and validity of these games are also tested.The second part chose 195 students from 4th grade to 6th grade as the subjects in a primary school in Hangzhou,and collect their executive function with online cognitive games.Based on these data,five classification models were built with five classification method.The performance of these models was tested,and the results show that the five models can accurately diagnose students with mathematical disability.Among the models,the K-nearest neighbor model has the best performance,the accuracy is 77.7%,and the recall rate is 96.5%.The whole diagnostic process was assisted by 3 simple trained staff.It only takes one hour to finish the diagnosis task of 40 students.Every participant showed high participate motivation.This study proposes a mathematical disability diagnosis method with the help of online cognitive games with an acceptable accuracy,less cost of manpower and time comparing to the traditional methods,and effectively maintain the student's motivation.This method is able to conducted in a wide range of mathematical disability diagnosis,and also provides new ideas for the diagnosis of other types of learning disability.