Method for evaluating dysarthria in children by optimizing and selecting acoustic parameters based on data mining technology

Abstract

The invention discloses a method for evaluating dysarthria in children by optimizing and selecting acoustic parameters based on a data mining technology. Objective acoustic features of a tested childare collected, extracted and analyzed; a LASSO regression function is adopted; a generalized linear model is fitted; variable selection is carried out; an integrated weight dependent variable is extracted; therefore, acoustic feature parameters having important reference significance on dysarthria in Putonghua-speaking children and the weights of all kinds of acoustic feature parameters are selected; and thus, the foundation can be laid for objective evaluation of dysarthria in children. A LASSO method is adopted for the first time in the invention; the correlation of 40 acoustic feature parameters and all kinds of dysarthria in children is analyzed; the weight acoustic features of all kinds of dysarthria in children are found; and the most contributive acoustic parameters for evaluating the dysarthria in Putonghua-speaking children are explored.

Description

Technical field [0001] The present invention belongs to the technical field of dysarthria evaluation, and specifically relates to a method for evaluating children's dysarthria based on data mining technology by selecting acoustic parameters. Background technique [0002] At present, the number of children with language barriers in China is increasing year by year, and the communication barriers caused by dysarthria seriously affect the children's return to society. Although the number of patients with dysarthria in my country is large, a 2016 survey by Lin Qiang and Lu Jianliang found that the current assessment methods cannot meet the therapist's needs for precise speech rehabilitation. Domestic rehabilitation departments and speech rehabilitation institutions still use subjective auditory evaluation and / or scales that require subjective judgment as the main evaluation methods, which lack objectivity and efficiency. [0003] In recent years, based on the rapid development of arti...

AI Technical Summary

Problems solved by technology

However, the data mining and neural network recognition of pathological speech are in their infancy. The pathological acoustic features and their changing rules extracted from children with dysarthria can objectively and quantitatively reflect the dysarthria in children, which is the premise of data mining and ANN fitting recognition. conditions, but the exact mechanism remains unclear

Images