Research On Quantitative Evaluation Method Of Bradykinesia In Parkinson Disease Based On Inertial Sensor

Parkinson’s disease is a neurodegenerative disease that occurs more frequently in elderly patients.Bradykinesia symptoms are common dyskinesia symptoms of Parkinson’s disease,which directly affect the daily life of all patients with Parkinso n’s disease.At present,the clinical evaluation method of bradykinesia symptoms is mainly scale evaluation,but scale evaluation often has the problems of greater subjectivity and the evaluation results may not be sufficient.Therefore,objective diagnosis and evaluation of bradykinesia symptoms are needed.This article mainly aims at the clinical quantification diagnosis and treatment needs of Parkinson’s disease,with wearable detection and evaluation based on inertial sensors as the direction,designing data collection schemes and experimental environments,collecting clinical information and motion data from patients with Parkinson’s disease and healthy control groups,from motion data Extract the characteristic parameters that can characterize bradykinesia,and build a quantitative evaluation machine learning model for Parkinson’s disease bradykinesia.The experimental results show that the classification accuracy of neural network multi-layer perceptron algorithm for Parkinson’s disease patients and normal subjects is more than 97%,and the classification accuracy of GBDT algorithm combined with decision tree model for quantitative classification of bradykinesia symptoms is more than 84%.The research in this paper validates the feasibility of the quantitative evaluation method of Parkinson’s disease bradykinesia based on inertial sensors.The extracted quantitative indicators and detection methods have certain reference value for the research and clinical assistant diagnosis and treatment in related fields.The following is the main research work of this article:(1)In-depth research on the application and research status of wearable devices in the quantitative evaluation of Parkinson’s disease with bradykinesia,and build a quantitative evaluation system based on multiple sensors and multiple motion tasks for bradykinesia symptoms.(2)Design the experimental data collection scheme with reference to the UPDRS scale,including designing experimental tasks,setting up experimental sites,screening experimental subjects,and collecting experimental data.(3)Design data processing algorithms,perform pre-processing on the collected raw data,perform time-domain and frequency-domain analysis,extract feature parameter sets that can be used for quantitative evaluation of bradykinesia.Establish data sets for quantitative analysis and assessment of bradykinesia symptoms based on feature parameter sets and clinical information of subjects,and standardize the data set.(4)Unbalanced data processing,to explore a variety of unbalanced data processing methods for the unbalanced problem of the data set.Finally,the resampling method of S mote+Tmoke link is used to resample the data set to obtain a balanced data set.(5)Based on the statistical analysis of the feature parameter set,the evaluation indicators that can be used for the quantitative evaluation of Parkinson’s bradykinesia are selected.The quantitative evaluation model of bradykinesia was established,and the performance of the model was evaluated.(6)According to the established quantitative evaluation model of bradykinesia,we design and develop an inertial sensor-based application system for quantitative evaluation of bradykinesia,which realizes the functions of basic patient information management,initial screening of Parkinson’s disease and quantitative evaluation of Parkinson’s disease.The system provides tools for clinical auxiliary diagnosis and treatment.