The Research On Analysis Of Parkinson's Disease Based On Speech Information Recognition Technology

Parkinson's disease is a very common neurological disease,in which patients with Parkinson's disease are more common in the elderly,Parkinson's disease has a wide range of disease,and the onset time is relatively long;however,the exact cause of Parkinson's disease is still It is not clear that medical personnel can only diagnose according to the patient's symptoms and clinical experience.Most patients with Parkinson's disease have different degrees of speech impairment in the early stage of the disease,so the study of the diagnosis of the severity of Parkinson's disease through voice information has become a research hotspot in this field in recent years.This article first describes the current status of Parkinson's disease and the significance of the diagnosis of Parkinson's disease based on voice information,and describes the future development of remote diagnosis and tracking of Parkinson's disease,as well as the current status of research on remote diagnosis of Parkinson's disease at home and abroad.Then introduce the relevant theoretical knowledge of speech analysis,as well as the machine learning algorithm used in this paper.Then,the laser spot experiment under different voice information of Parkinson's disease was carried out.Different Parkinson's speech signals will produce different spots,and the more severe the patient's speech signal,the more complicated the spot light,which proves that the voice information and the disease are serious.The degree is essentially linked,and it is feasible to diagnose Parkinson's disease through voice information.Secondly,UPDRS is a unified clinical reference standard for assessing the severity of Parkinson's disease symptoms.In this paper,speech signal processing algorithm is introduced to analyze the speech information of Parkinson's disease patients,and the physical meaning of each speech feature is analyzed.After processing,machine learning algorithm is used to map the speech feature to UPDRS to realize the remote diagnosis of Parkinson's disease.In conclusion,clinical staff can prompt patients' actions based on the estimated UPDRS.This paper uses three kinds of model decision tree,random forest and support vector machine.Then,cross-validation is used to evaluate the performance of the model,and practical MAE is used to compare the performance of the three models.The correlation strength between each speech feature and UPDRS is also analyzed.The research and diagnosis of Parkinson's disease based on voice information has important academic research significance and wide application prospects.The remote monitoring ofParkinson's disease diagnosis method introduced in this paper is much better than the traditional Parkinson's disease diagnosis method.