Research And System Development Of Parkinson's Disease Detection Based On Speech Spectrogram

Parkinson's disease is the second largest neurological disorder after Alzheimer's disease.Its clinical manifestations mainly include resting tremor,bradykinesia,muscle rigidity and postural gait disorders,while patients may have depression,constipation and sleep disorders And other non-motor symptoms,seriously affecting the quality of life of patients.Early detection of its symptoms and timely treatment can help patients significantly improve their quality of life.Therefore,building a Parkinson's disease early diagnosis system is of great significance.The related research shows that early diagnosis of Parkinson's disease can be achieved by using the voice of Parkinson's disease patients and machine learning methods.On this basis,this paper discusses the characteristics of speech,spectrogram of speech,convolutional neural network and typical machine learning platform under big data,and presents a detection method of Parkinson's disease based on spectrogram of speech.The deep learning model in this approach is trained by using the Spark-based DeepLearning4 J toolkit.The Experiment have shown that using voice spectrogram combined with convolutional neural network to detect the symptom of Parkinson's disease can reach high degree of accuracy.At the same time,this paper develops Parkinson's disease detection system,the system has four modules,includes user management,voice upload,Parkinson's disease detection,database module.After the user logs in,the user can upload a continuous vowel / a / or / o / audio to the server in the user center After the audio being processed by the server,the server will feed back the result to the user.This system is user-friendly,and is easy to operate,and the detection speed is fast,what's more the detection accuracy of Parkinson's disease is high.