Research And Implementation Of Aided Prediction Technology Of Disease Based On Stream Data

With rising living standards,people's attention on health increases.The outbreak,development and dissemination of a variety of diseases has close relationship with meteorological conditions.There is a positive significance for the prevention and early warning of weather sensitive diseases of analyzing the impact of climate change on the health of the population.However,with the growing number of meteorological factors,how to efficiently obtain potentially valuable information for predicting the disease has caused widespread concern of scholars and research institutions.At present,domestic and international research on the impact of meteorological factors for the disease are multiple regression analysis,artificial neural networks and traditional decision tree methods,depending on the characteristics of different weather environments and disease data analysis of the relationship between climate change and disease.But these methods have higher requirements on the initial data set,ignoring the close correlation and similarity between meteorological factors of different dates and models are built on the limited historical data set,without taking into account the influence of the latest meteorological information on the disease.In view of the above problems,processes meteorological data with time series as data stream,and this paper puts forward two prediction algorithms of weather-sensitive diseases.Disease prediction model of JacUOD algorithm based on data stream first calculates the similarity with the target date for the historical date meteorological factors in the sliding window by filtering data sets,select the highest similarity Top-N historical dates' meteorological data for patients' numerical prediction of the target date of disease,make full use of close relationship of meteorological factors and real-time meteorological data;the decision tree algorithm based on data stream uses the time sliding window technology to extract the latest meteorological data,combined with the decision tree algorithm to establish disease prediction model,selects the latest meteorological data for disease prediction in dynamic data sets.The above two methods are used to predict the weather-sensitive diseases,which can improve the accuracy of the algorithm.Combined with the above proposed prediction algorithm,completed the "weather and health" APP prototype design and development and achieve the common forecast of nine kinds of weather sensitive diseases in Beijing city.Experiments show that the two proposed disease prediction algorithms make full use of the latest meteorological data and the close relationship of meteorological factors at the same time,greatly enhance the accuracy of disease prediction.