Study On The Relationship Between Meteorological Factors And Several Chronic Diseases Based On GAM Model

Meteorological factors have an important impact on the onset of chronic diseases.Using scientific methods to study health impact factors related to chronic disease episodes can provide valuable information for public health burdens,help public health and environmental departments to plan properly,and disseminate climate change to the public in a timely manner.Response.This dissertation takes the meteorological factors of representative regions of four different climates in Gansu Province and the data of several chronic diseases(pulmonary heart disease and hypertension)in the same period as the research object.Firstly,the GAM model is optimized,and the DLNM model is combined to study the exposure-response relationship and hysteresis effect.Then,a neural network compensated gray periodic epitaxy algorithm is proposed to solve the periodic fluctuation of the time series.The problem is to improve the prediction accuracy of the outpatient volume.The main work of the thesis is:First,using descriptive statistical analysis,correlation analysis and principal component analysis to analyze the daily meteorological data and contemporaneous period of the four representative regions of Gansu Province(Liangzhou District,Baiyin,Qingcheng and Chengxian Districts)from 2014 to 2016.The preliminary analysis of the number of outpatients of chronic diseases(pulmonary heart disease and hypertension)showed major meteorological factors affecting chronic diseases.Then,the smoothness optimization and co-curve optimization of the GAM model were carried out.Combined with the DLNM model,the relationship model between meteorological factors and chronic disease clinics was established and applied to four representative regions.Both the GCV and K-index indicators of the optimized GAM model indicate that the established model has sufficient smoothness and the degree of common curve is reduced,and the model effect is better.The experimental results show that the four regions have different regional and climatic differences,so the temperature is two.The degree of influence of chronic diseases varies,but the relationship between temperature and lag time on pulmonary heart disease and hypertension is nonlinear,showing a slow hysteresis effect on both low temperature and high temperature,and the hysteresis effect map is mostly “U”type.Finally,aiming at the problem of low accuracy of outpatient volume prediction,a neural network compensation gray period epitaxy algorithm is proposed,which considers the seasonal cycle volatility of outpatient volume.By comparing the proposed algorithm with the GM(1,1)model and the gray periodic epitaxial model,the results show that the average relative error of the prediction results of the GM(1,1)model is 6.6487%,and the gray periodicepitaxial model predicts the results.The relative error is 6.1211%,and the average relative error of BP neural network compensation period extension algorithm is 1.1892%,and the three precision evaluation indexes of this algorithm are more than 50% lower than the other two.The algorithm was applied to the number of pulmonary heart disease and hypertension in the four representative regions of Gansu Province,and the prediction analysis was carried out.Combining the prediction results of this method with the previous exposure-response relationship and hysteresis effect can provide an effective scientific basis for government departments and epidemiological prevention and control measures.