Research On The Diagnosis System Of Parkinson’s Disease Based On Mobile Terminal

Parkinson’s disease is a degenerative disease of the neurological system and the number of patients is increasing year by year and showing a trend of younger people.At present,the diagnosis of the disease mainly relies on the clinical experience of medical experts,since it cannot provide an objective basis,the diagnosis is too subjective,and it is easy to cause misjudgments.To diagnose Parkinson’s disease more accurately and objectively,this paper proposes three diagnostic methods for Parkinson’s disease based on persistent vowel signal,hand-drawn spiral diagram,and multimodal fusion for Parkinson’s patients with speech disorder and hand movement disorder.It provides patients with remote self-examination and clinical assistance for physician evaluation and treatment.This paper mainly focuses on the recognition of speech disorders and hand movement disorders in Parkinson’s patients.The main contents are as follows:(1)Aiming at the early speech impairment of Parkinson’s patients,a speech feature fusion algorithm based on a convolutional gated recurrent network is proposed to extract the pathological features in the patient’s continuous vowel signal to diagnose Parkinson’s disease.Acoustic features and spectrogram features can better capture the dynamic pathological information of the speech.After comparative experiments,the algorithm can achieve better diagnostic results than traditional speech feature extraction combined with machine learning algorithms.(2)Aiming at the hand movement disorder of Parkinson’s patients,a transfer learning algorithm based on multi-scale convolution is proposed.Through the superposition of different scale convolutions,local features and global features are fused to improve the network’s ability to understand pathological features in handdrawn images.In the case of a few data samples,the idea of transfer learning is introduced,and the model trained in advance on a large data set is used to fine-tune the target sample.After experimental verification,the algorithms can achieve better diagnostic results compared with the traditional hand-drawing diagnosis method.(3)Aiming at the variability and complexity of Parkinson’s disease progression,it is difficult to achieve an accurate diagnosis of the disease with the characteristics of a single modality.A diagnostic method based on the multimodal fusion of speech and hand-drawn spirals is proposed,which simultaneously evaluates patients’ language impairment and hand movement disorders,the influence of different feature fusion methods on the diagnosis results is analyzed.Experiments verify that the multi-modal diagnostic scheme of decision-level fusion of acoustic features and hand-painted spirals can achieve better diagnostic results than a single modality.Finally,combined with the three deep learning based Parkinson’s diagnosis methods proposed in this paper,a mobile terminal platform for disease diagnosis based on speech and hand-drawing is designed and implemented.It can help reduce patient visits to the hospital for clinical examinations and reduce the workload for clinicians.