人工神经网络在气象因素预测急症发病中的应用

Application of Neural Network in Forecasting Emergencies Caused by Meteorological Factors

目的:探讨各种急症发病与气象条件的关系,建立基于人工神经网络的医疗气象预报模型,为预防和控制各种急症的发病提供科学参考。方法:收集济南市急救中心2007～2008年的急诊病例和同期天气资料,利用SAS9.0统计软件进行气象因素与各种急症发病人数的相关分析,利用Matlab7.0软件构建急症发病的BP人工神经网络预测模型,并对网络进行评价。结果:气象因素及其变化与各种急症的发病有密切关系。根据建立的人工神经网络模型预测显示,除CO中毒预测准确率较低外(46%),其余各类急症的预测准确率为76%～89%。结论:基于BP人工神经网络的急症医疗气象模型有较好的预测效果,具有进一步研究的价值。

Objective： By analyzing the relationship between the meteorological factors and the outbreak of emergencies, an artificial neutral network model was established in order to make the medical-meteorological forecast and to reduce and prevent the emergencies. Methods： The data of emergency cases and meteorological factors within the same time in 2007 and 2008 in Ji＇nan, China were collected and analyzed by using SAS9.0. The back-propagation （BP） artificial neutral network model was built by using Matlab7.0. Results： A close relationship exists between the meteorological factors as well as their changes and emergencies. The results of forecast show high accuracy rate of each emergency （76% - 89%） except CO poisoning （46%）. Conclusions： This emergencies medical-meteorological forecasting model based on BP neutral network has good prediction effect and the value of further research.