Nowadays,in every aspect and field of people’s production and life,the shadow of artificial intelligence is everywhere.Artificial intelligence is gradually changing our production mode and living habits.Machine learning is the main branch of artificial intelligence.It can automatically mine learning patterns from data,so as to help people make more scientific and objective decisions,and reduce the error rate of human decision-making.Intelligent diagnosis of medical diseases is an important application scenario of machine learning methods.It is of great significance to build a powerful,efficient and applicable intelligent decision-making model to provide scientific and reasonable decision-making assistance for disease diagnosis.In this paper,machine learning methods are studied around a variety of disease diagnosis problems,so as to provide decision-making reference and to construct intelligent diagnosis models for auxiliary diagnosis of diseases such as skin diseases,pleural effusion and kidney stones.In this paper,machine learning methods such as extreme learning machine and kernel extreme learning machine are mainly studied,and on this basis,how to improve the existing methods to achieve better prediction effect in disease diagnosis is deeply analyzed.In this paper,we propose the skin disease diagnosis model based on the optimization of the extreme learning machine by the decentralized feeding sine cosine algorithm,the pleural effusion diagnosis model based on the optimization of the extreme learning machine by the chaos differential evolution salp swarm algorithm,and the kidney stone diagnosis model based on the optimization of the kernel extreme learning machine by the elite chaos slime mold algorithm.The main research contents of this paper are as follows:(1)This paper expounds the research status of machine learning-based diagnosis of related diseases and the extreme learning machine based on intelligent optimization method,analyzes the shortcomings of the existing research,and makes a tentative plan for the research work in this paper.At the same time,common machine learning methods and intelligent optimization algorithms are briefly introduced,with emphasis on the extreme learning machine,kernel extreme learning machine and sine cosine algorithm,salp swarm algorithm,slime mold algorithm.As theoretical support,these related knowledge provides theoretical basis for the follow-up work of this paper.(2)In order to improve the accuracy of skin disease diagnosis,a extreme learning machine model(DFSSCA-ELM)based on the optimization of the sine cosine algorithm of decentralized foraging was established to diagnose erythema squamous skin disease.Firstly,a sine cosine optimization algorithm(DFSSCA)based on decentralized foraging was proposed by introducing the decentralized foraging mechanism.The CEC2017 benchmark function test set was used to evaluate the optimization performance of DFSSCA,and the performance was compared and analyzed with 8 swarm intelligence algorithms with superior performance.Then,based on DFSSCA method,the weight and threshold of the extreme learning machine were optimized,so as to propose a new prediction model(DFSSCA-ELM),which was used to diagnose and predict erythema squamous skin disease.Meanwhile,it was compared with 5 machine learning models and the evaluation index was used to measure the overall performance of the model.The experimental results show that the DFSSCA-ELM model has achieved significant improvement in the evaluation indicators,indicating that the decentralized foraging mechanism can effectively improve the predictive ability of the model.(3)In order to improve the accuracy of pleural effusion diagnosis,this paper proposed an improved extreme learning machine model(CDESSA-ELM)based on chaotic differential evolutionary salps swarm algorithm to diagnose pleural effusion.Chaos initialization mechanism and differential evolution mechanism are respectively used to improve the global and local searching ability of salp swarm algorithm,and an improved salp swarm algorithm is formed(CDESSA).First,the CEC2014 function test set was used to evaluate the optimal performance of CDESSA and compare it with 7 improved swarm intelligence algorithms.Then,CDESSA algorithm was used to optimize the parameters of the extreme learning machine model,so as to propose a new prediction model(CDESSA-ELM),which was used to predict the pleural effusion disease.Meanwhile,a comparative study was carried out between the model and four machine learning models,and the overall performance of the model was measured through evaluation indexes.The experimental results show that the CDESSA-ELM model achieves significant improvement on all evaluation indicators,indicating that the chaotic initial mechanism and differential evolution mechanism can effectively improve the predictive ability of the model.(4)In order to improve the accuracy of kidney stone diagnosis,a kernel extreme learning machine model(ECSMA-KELM)based on elite chaotic slime mold algorithm was established to diagnose kidney stone disease.First through a combination of elite mechanism and chaos mechanism put forward an slime molds improved optimization algorithm(ECSMA),makes the global search ability and local search ability to achieve a more stable state.Using 23 benchmark functions and 8 composite functions from CEC2014 function test set to evaluate the optimal performance of ECSMA algorithm,and compare it with the 12 improved swarm intelligence algorithm.Then,the key parameters of kernel extreme learning machine were optimized based on the ECSMA algorithm,and an optimal model(ECSMA-KELM)was constructed for the diagnosis of kidney stones.Meanwhile,the model was compared with 5 machine learning models.The experimental results show that the ECSMA-KELM model can significantly improve all the evaluation indexes,which indicates that the elitist mechanism and chaos mechanism can effectively improve the prediction ability of the model. |