Prediction Of Cerebrovascular Morbidity Based On Improved Extreme Learning Machine

The occurrence of cerebral vascular disease has a great randomness and uncertainty,affected by many factors.The meteorological factors are very important.The prediction model of cerebrovascular disease incidence based on meteorological data can be used to provide reference for prevention and treatment of cerebrovascular diseases.Based on the above situation,the following research are carried:1.Firstly we understand the current situation of cerebrovascular disease,and explored and summarized the method of predicting the number of patients with cerebrovascular disease and the influencing factors.On the basis of the present situation of cerebrovascular disease and the shortage of prediction methods,the research content of this subject is determined.2.According to the characteristics of the problem,Propose the prediction method of the number of cerebrovascular diseases by using the limit learning machine.Extreme learning machine algorithm compared with the traditional machine learning algorithm has the superiority,but also has some limitations.In order to improve the kernel function approximation ability,propose a combination of Gauss kernel function kernel and wavelet kernel to structure combination kernel extreme learning machine,and improve the extreme learning machine in processing multidimensional data accuracy.3.In order to improve the performance of the combinatorial core limit learning machine,the particle swarm optimization(PSO)algorithm is proposed to optimize the parameters of the nuclear limit learning machine,and the inverse learning initialization method and Gauss mutation operation are introduced to improve the optimization ability of the particle swarm optimization algorithm.Finally,MATLAB is used to carry out simulation experiments.Experiments show that the improved particle swarm optimization(MPSO)algorithm has better search ability.4.Optimize the kernel parameters of the combined kernel learning machine by using the ability of MPSO global search.At the same time,according to themeteorological data and the correlation of the number of cerebrovascular disease,the prediction model of the number of cerebrovascular diseases was established based on the MPSO optimized combination nuclear limit learning machine,and the number of cerebrovascular diseases was predicted every seven days by using the meteorological data of the past seven days and the next seven days.Take the data of Zhengzhou city from 2015-2016 as an example.Compared with the PSO-ELM algorithm,SVM algorithm,BP neural network algorithm,the experimental results show that the optimal combination of MPSO nuclear ELM algorithm has significant performance.The algorithm can better solve the incidence prediction problem,and the prediction effect is better than model.The number of cerebrovascular disease prediction,prevention and monitoring can be achieved on the incidence of cerebral vascular promptly and timely,and provide strong information support for the health management and scientific decision-making.With the gradual accumulation of chronic disease data,this model can be used for early warning and forecasting of other parts of the province.