Application And Research Of Machine Learning Algorithm In Disease Diagnosis

With the rapid development of economy,people's pace of life is accelerating.The incidence rate of diseases caused by various bad habits is also increasing.But due to the limitation of medical resources and the complexity of disease diagnosis,the existing medical conditions are hard to meet the increasing demand for medical treatment.In addition,with the development of medical information technology,medical institutions have accumulated a lot of data.Therefore,it is of great significance to study how to use the existing medical data to serve the clinical work,so as to reduce the burden of medical workers and improve their work efficiency.In this paper,machine learning algorithm is used to build disease diagnosis model,which can provide a powerful reference for actual disease diagnosis,reduce the impact of various subjective factors on decision-making results,decrease the probability of misdiagnosis,and improve the efficiency of disease diagnosis.The main research contents and results are as follows:(1)In order to reduce the influence of irrelevant features in medical data on model performance,a feature selection method based on improved grasshopper optimization algorithm is proposed.Firstly,based on the traditional grasshopper optimization algorithm(GOA),nonlinear decreasing coefficient,adaptive position weight and limit threshold are introduced to improve GOA.The effectiveness and superiority of the improved algorithm(IGOA)are proved by benchmark function.Secondly,IGOA is used to search the feature subset of data set,and a feature selection method based on IGOA algorithm(IGOA-FS)is proposed.Finally,the experimental results show that IGOA-FS can effectively select datasets features,reduce data dimension and improve model performance.(2)In order to improve the classification performance of the model,an SVM disease diagnosis model based on the improved whale algorithm is proposed.Firstly,adaptive probability threshold,individual position update weight and preselection niche technology are introduced to improve the traditional whale optimization algorithm(WOA),and an improved algorithm(APN-WOA)with higher precision and faster convergence speed is proposed.Secondly,APN-WOA is used to optimize the penalty coefficient C and kernel function bandwidth ? of SVM,and a SVM disease diagnosis model based on APN-WOA algorithm(APN-WOA-SVM)is proposed.Finally,the performance of APN-WOA-SVM is tested on different disease datasets.The experimental results show that APN-WOA-SVM has higher classification accuracy than other comparison algorithm.(3)In order to complete the optimization of model parameters and the feature selection ofdatasets synchronously,a disease diagnosis model(FS-SVM)is proposed,which combines the IGOA-FS feature selection method with APN-WOA-SVM model.Firstly,the performance of IGOA and APN-WOA is compared.Secondly,APN-WOA with better performance is selected to optimize the feature subset of datasets,parameter C and? of SVM.Finally,the performance of FS-SVM is tested on different disease datasets.The experimental results show that the classification performance of FS-SVM is significantly improved compared with APN-WOA-SVM,and compared with other algorithms,the classification performance and feature selection result of FS-SVM still have some advantages.