Research On Method And Technology Of Medical Meteorological Service Based On Large Data

With the continuous development of information technology, the large amount of data generated in the field of medicine and meteorology has been accumulated, meanwhile, new data are still being produced continuously, which takes on the feature of big data. Undoubtedly, it lays on the significant data foundation of Medical meteorology, a new cross subject, and brings power and activity for further researching the impact of changes in weather and climate on human health. Therefore, the purpose of this paper is to introduce the idea of big data to medical meteorological service,to present the framework of medical meteorological service in the era of big data, and to research how the method and technology of large data apply in medical meteorology service to promote the rapid development of medical meteorological.In this Dissertation, the framework of medical meteorological service based on big data is tried to be established, which is divided into four layers from bottom to top: the Data Source Layer(DSL), the Integration Layer(IL), the big Data Platform Layer(DPL) and Application Layer(AL). This framework can provide a series of solutions, from the data acquisition, storage, pre-processing, the establish and application of forecasting model and a series of solutions for application requirements. Secondly, according to the characteristics of medical meteorological data, a data storage framework based on big data is proposed. The framework uses advanced technology, distributed technology and web crawling technology to obtain the structured and unstructured medical, meteorological and air pollution monitor data of different storage locations. The storage frame also constructs the medical meteorological data storehouse, which is used to carry on Quality Control and storing acquired relative data in time.At the same time, based on medical data and corresponding meteorological and air pollution index data in a Tertiary hospital in Lanzhou, the paper carries on the construction trial of data storehouse instance. Furthermore, we compared existing pre-processing technology according to the character of large data of medical meteorology, carried on data transformation and prediction factor’s selecting, and the results show that in the absence of data processing(data transformation), the accuracy rate for the number of patients with high blood pressure, upper respiratory tract infection and diabetes mellitus changes were: 4.20%, 2.91% and 2.70%, the accuracy of the data pre-processed forecast rose to : 25.50%, 41.50% and 19.50%. Obviously the accuracy of the forecast was significantly improved. After the feature selection, the meteorology forecast factors of three above kinds of diseases were reduced to 9, 10, 10, and in the same server, the BP three kinds of diseases were reduced from before 19.45s、27.05s、31.04 s to 11.20 s, 6.39 s and 27.05 s. It can be seen from above the accuracy and the modeling time is significantly improved by the technology pre-processing of the medical meteorological data.Furthermore, the establishment of forecasting model is based on the large data of medical Meteorological Service forecast model construction of real demand. In the another research focus, by comparing the various forecasting model algorithm which has been used in meteorology service forecasting, it proposed a new method of medical meteorology service forecast model under a large data era-genetic algorithm of nonlinear programming supporting vector regression algorithm, and use related data which has bee stored to verify, the results show that:(1) The prediction accuracy of the algorithm is significantly improved. Taking the above three examples, the accuracy of prediction is 62%, 98.75 and 75.50%,(2) Establishing forecasting model by using the algorithm, the modeling time is better than the other algorithms. The modeling time of the above three diseases are reduced to: 4.57 s, 4.23 s and 3.20 s.(3) The algorithm is easy to be realized in the application, and in the article the algorithm is realized by using the library functions and related technology already existing in the.NET 2008. The storage space is small, the program is running fast, wand it is suitable for the real-time prediction of medical meteorology service.Finally, the article tries to develop the medical meteorological prediction and software platform of service system, in addition to the research of WEB, but also developing the current popular Mobile Internet, including: mobile phone APP Based on Android system and IOS system, and developing of micro channel public number. The software platform not only shows the users the results of medical meteorological forecast in a variety of forms through the friendly interface, but also by locating the users, push the location of the meteorological and environment information and disease-preventing knowledge to remind the user to prevent timely. This software platform has the characteristics of big data era. It has a good ability of early warning for the related disease patients group. At present, it has been tried and achieved good service effects.