Cooperative Optimization Of Gait Feature And Machine Learning Model Parameters In Patients With Parkinson’s Disease

As China’s aging situation is highlighted,the high incidence of Parkinson’s disease in the elderly has become a huge problem for families and society.In the clinic,doctors diagnose patients with the help of scales and empirical judgments,but there is a risk of misdiagnosis.Therefore,in order to assist doctors in the clinical diagnosis of patients with Parkinson’s disease,this paper uses genetic algorithm and ant colony algorithm to collaboratively optimize the gait characteristics and support vector machine parameters of patients,and provide auxiliary diagnosis based on patient gait characteristics.The following is the main work of this paper:(1)Using genetic algorithm to study the collaborative optimization of patient gait characteristics and support vector machine parameters,the patient features constitute the chromosome in the algorithm,and the support vector machine parameters are also incorporated into it to explore the synergistic optimization of the two.In this paper,some feature selection algorithms are selected for the results of collaborative optimization,and only the genetic algorithm is used for feature selection as a control group,the experimental results show that the use of genetic algorithm for collaborative optimization improves the classification accuracy of the model compared with the control group algorithm.(2)Using ant colony algorithm to study the collaborative optimization of patient gait characteristics and support vector machine parameters,use the feature as the node on the ant search path,and the support vector machine parameters are searched together as part of the feature,and simultaneously complete the feature selection process.Selection of parameters.Compared with the control group using only ant colony algorithm for feature selection and other selection algorithms,the experimental results show that the ant colony algorithm has obvious advantages in collaborative optimization,with high classification accuracy,and the average accuracy rate is 85.77%.