Research On Stroke Burden Prediction Model In China Based On GLM Method

With the serious ageing of our population and changes in the lifestyle of the population,the rising burden of stroke is seriously affecting the lives and property of the population.Disease burden indicators such as prevalence,incidence,deaths and disability life-adjusted years have become valid public health indicators for assessing the quality of survival and life expectancy loss of stroke patients,and an important decision basis for governments to formulate health policies and allocate health resources.This paper analyses the study of a model for predicting the burden of stroke in China based on the GLM approach as follows.First,descriptive research.Preliminary description of the prevalence,incidence,deaths,and disability life-adjustment years of stroke in my country from 1990 to 2019,as well as the population change and the development of per capita GDP during this time period,and preliminary exploration of the disease burden of stroke(Incidence,Prevalence,Deaths and disability-life lost years),the number of elderly people over 70 years old and the development trend of economic indicators represented by per capita GDP in the past 30 years and the possible reasons for the change.Secondly,Proposing a model for predicting the burden of stroke in China based on the GLM approach.Due to the stringent data quality requirements of the Lee-Carter method and the relatively simple model structure involving only one time-effect variable,which is not an accurate fit to historical data,the traditional GBD generalised linear model differs from the Lee-Carter method in that it considers the relationship between a burden indicator and a set of variables.The traditional GLM generalised linear model differs from the Lee-Carter approach in that it considers the relationship between burden indicators and a set of variables,allowing for the effects of advances in medical care and technology,but it is only applicable to older age groups and is less effective in fitting other age groups.Therefore,this paper improves the generalised linear model by introducing an innovative population exposure factor to account for the differences in stroke burden indicators between different age groups and to further improve the fit and prediction accuracy of the model.In order to verify the accuracy and reliability of the model,the data from 1990 to 2019 were split into training data and accuracy validation data,with the data from 1990 to 2009 being the training data and the prediction models differentiated by country,sex and age group.The model parameters are then tested for significance and predicted for the relevant groups from 2010 to 2019.The error between the results and the real data is tested using root mean square error to select the best-fit model.The experimental results show that the GLM-based stroke burden prediction model in China is more accurate than the Lee-Carter model in traditional epidemiology,has a simple and clear structure,is well adapted to most age groups,and can effectively predict future changes in disease burden indicators.Finally,the burden of stroke in China is projected for the next two decades and the reasons for the change in outcome trends are discussed.The best-fit model was selected based on the minimum root mean squared error from the internal validation done above,and the predicted model was constructed based on historical empirical data for China from 1990 to2019.The reasons for the differences are discussed and recommendations for stroke prevention based on the current situation in China are made.This paper proposes an effective method for predicting the burden of stroke based on the GBD generalized linear model.It gives the trend of future stroke changes from the multi-scale perspective of country,gender and age group,in order to update the country’s prevention and treatment measures in the field of stroke.And provide certain reference for the formulation of related policies.