Research On Prediction Of Hepatitis A Incidence Based On Improved Extreme Learning Machin

Since Hepatitis A has a long preclinical period and is highly contagious,the research on the prediction of the number of Hepatitis A incidence have an important reference value for the national control of the incidence trend of Hepatitis A disease,the formulation of prevention and control measures.Due to many factors affecting the number of patients with Hepatitis A,the sequence is nonlinear and non-stable,and the accuracy of traditional prediction methods is not high.In recent years,the swarm intelligence algorithm has been applied in many fields,which has the advantages of strong global optimization characteristics and few adjustment parameters.Therefore,this paper uses the swarm intelligence algorithm to optimize the parameters of ELM network structure and predict the number of patients with Hepatitis A by combining the idea of ”decomposition first and then combination”.The specific work of this paper is as follows:(1)Aiming at BOA slow searching speed and easy to fall into local optimization,an improved butterfly optimization algorithm(IBOA)was proposed.In the initial stage,IBOA uses Cubic chaotic mapping to generate individual positions in the population to enhance the diversity of the population.In the optimization stage,IBOA uses dynamic weight to enhance the global optimization ability and accelerate the convergence speed of individuals,and uses refraction reverse learning to further enhance the ability of individuals to jump out of local optimum.In order to verify the optimization speed and accuracy of IBOA,IBOA was tested on multiple benchmark functions with 30 and100 dimensions respectively.The results show that IBOA has faster convergence speed and more stable optimization ability compared with BOA,HPSOBOA,PSO,WOA and MPA.(2)Aiming at the deficiency of randomly given input weights and thresholds of ELM,this paper optimized the input weights and threshold parameters of ELM by the improved butterfly optimization algorithm proposed in this paper,so as to establish the IBOA-ELM prediction model.The monthly incidence of Hepatitis A in China from 2007 to 2019 was selected for model training,and the monthly incidence data of Hepatitis A in 2020 was used to verify the performance of the model.The prediction results showed that the ELM parameters sought by IBOA with strong global searching ability improved the prediction accuracy of the test set.The MAPE predicted by the IBOA-ELM model was 5.99%,which was higher than that of the BOA-ELM,ELM and ARIMA models.(3)Aiming at the nonlinear and random fluctuation of the number of patients with Hepatitis A,an ensemble empirical mode decomposition(EEMD)method combined with IBOA-ELM is proposed in this paper.EEMD method decomposes the sequence into multiple components with different time scale distribution rules according to the intrinsic characteristics of the sequence,thus reducing the non-stationarity and complexity of the sequence is conducive to the establishment of the prediction model.After combining the ARIMA,ELM,BOA-ELM and IBOA-ELM models with EEMD,the prediction accuracy was improved.Among them,the EEMD-IBOA-ELM model can capture the nonlinear characteristics of sequence more accurately,has the highest prediction accuracy is the among all models,reducing the prediction predicted MAPE to 3.14%.