Establishment And Preliminary Application Of Early Warning Technologies For Respiratory Diseases In Shanghai

In recent years, the prevention and control of infectious diseases in Shanghai has made great achievements. There is a significant decline in the incidence of infectious diseases. But, new respiratory infectious diseases continue to emerge, while the old respiratory infectious diseases rage again, which make a major threat to people’s health and the social economy. The experiences of SARS, avian flu, and H1N1 influenza indicate that good forecasting and early warning of infectious disease and taking response measues ahead is an important guarantee of the prevention and treatment of respiratory diseases.Infectious disease prediction is a statistical analysis method to combine mathematics and epidemiology of infectious diseases. It is based on the occurrence and development of the law and the relevant factors of infectious diseases through the analysis and application of mathematical models to determine methods for infectious disease trends and changes in the level of development may occur judge. Forecasts provide the basis for the development of infectious disease prevention and response strategies to control infectious diseases.Currently the use of predictive models of infectious diseases is very extensive. ARIMA time series analysis model is an important and fundamental models. It is widely used to predict the incidence of infectious diseases. Especially for seasonal changes in time series modeling approach has been widely used in the medical field. ARIMA model is suitable for short-term forecasting of infectious diseases, with practical, high accuracy characteristics. The artificial neural network has a unique parallel architecture, adaptive, self-organizing, associative memory, a strong fault tolerance and anti-conversion and other characteristics, as well as unique information processing method, it can adapt to the complexity of some infectious diseases nonlinearity and uncertainty, but also the practical application and made remarkable achievements. BP neural network model and generalized regression neural network model uses the most. Combination of various forecasting models, is the development direction of infectious disease forecasting methods.In recent years, the number of influenza monitoring stations in Shanghai has gradually expanded. Reporting and monitoring infectious respiratory disease accumulated data, which gives the establishment of respiratory disease forecasting techniques provide great support.Meanwhile, the city’s Municipal CDC-county CDC-Community Health Center has been working with more mature network operating mode。 It is possible to carry out the technical basis for the use of predictive models of respiratory disease monitoring and early warning. Objective:For.the prevalence of acute respiratory infections in Shanghai, we choose a representative infectious diseases of respiratory data (influenza-like illness, scarlet fever), focusing on a variety of techniques to build predictive models of early warning, monitoring and early warning platform to build respiratory diseases, conduct preliminary application respiratory diseases early warning attempt to improve the ability of community districts and integrated control of infectious diseases, for large cities to establish a comprehensive monitoring and early warning system for infectious diseases, to provide a reference. Methods:1, According to the 2001 to 2012 data of Shanghai Class A and B infectious diseases, recalling the city in recent years, the prevalence of respiratory diseases.2, To establish the incidence of infectious diseases of trend forecasting model, choose a representative of infectious diseases of the respiratory tract, application ARIMA model and ARIMA and neural network model, choose January 2004 to April 2006 (123 weeks ILI surveillance data) to conduct ARIMA modeling analysis. And with dual linear and nonlinear characteristics according to infectious disease incidence data, select a monthly incidence of scarlet fever in Shanghai during the period 2005 to 2012 a total of 90 months, try to use a combination of ARIMA and ANN models performed monthly report the incidence of scarlet fever analysis and forecasting.3, The initial establishment and application of early warning platform. Initial respiratory disease monitoring and early warning system set up in the urban centers Disease Control and Prevention to carry out the application, the mathematical model implanted in early warning platform, continues to collect counties and communities such as the number of reported cases of influenza-like respiratory data, suggesting possible aggregation and outbreaks.Results:1、Since 2001, the incidence of respiratory diseases are among 37-60/10 thousand. During this time, there still occurs infectious diseases caused by respiratory infection that made greater impact. The emergence of new-onset respiratory diseases, such as new SARS, H1N1 flu and so on. Scarlet fever had been effectively controlled, but the incidence has another rise in recent years.2, The results show that ILI data with seasonal and non-seasonal characteristics of both. Using ARIMA (1,0,0) (1,1,0) 26 product models ILI incidence data fit very well. We use the information to build predictive models 1-114 week to predict the incidence of influenza-like illness of 115-123 weeks. All actual values are in 95% confidence interval. Predicted value was 100%.Scarlet fever incidence data month study showed that the best model to predict the effect of the first month is the ARIMA-GRNN combination model, the relative error is only 0.170; optimal model to predict the effect of the previous three months combined model is ARIMA-BPNN, MER 0.157; Three months after the best model to predict the effect of a simple ARIMA model, MER 0.078; optimal model to predict the overall effect is a combination of ARIMA-BPNN model, MER 0.131.It further confirmed that a combination of mathematical models can be used to analyze respiratory infections and forecasting.3, The use of early warning and forecast information system. Now they are a daily reported through network, then making real-time warning. Early warning and rapid disposal capacity of districts and communities has increased。 It also can help to detect outbreaks seedling。 Efficiency of early warning system is still to be confirmed by further studies.Conclusion: Through the recent epidemic of respiratory diseases in Shanghai review, we summarize the main features of respiratory diseases in recent years in Shanghai. The importance of early warning of infectious diseases has been identified. We chose representatives diseases respiratory diseases (influenza-like illness and scarlet fever cases) to build predictive Model. We try to establish ARIMA model and combined model of ARIMA and artificial neural networks. It provides respiratory viable technology to the establishment of early warning and forecast system. We initially established early warning platform of respiratory diseases. By monitoring the data and using predictive mathematical models, we attempt to provide a reference for effective early warning of diseases.