多种气象统计模型对比研究——以气象敏感性疾病脑卒中预报为例  

Comparison of various meteorological statistical forecasting models-Taking causing-stroke weather forecasting as an example

作　　者：刘博[1,2] 党冰[2,3] 张楠[1] 王式功[2,4] 尹岭[5] 张晓云[6] 黎檀实[5] 卢震华 LIU Bo1,2,DANG Bing2,3 ,ZHANG Nan1, WANG Shi-gong2,4, YIN Ling5 ,ZHANG Xiao-yun6 , LI Tan-shi5, LU Zhen-hua7(1. Tianjin Meteorological Observatory, Tianjin 300074, China; 2. College of Atmosphere Science, Lanzhou University, Key Laboratory of Arid Climatic Change and Disaster Mitigation of Gansu Province, Lanzhou 730000, China; 3. Beijing Municipal Climate Center, Beijing 100089, China; 4. School of Atmospheric Science, Chengdu University of Information Technology, Chengdu 610103, China; 5. General Hospital of PLA, Beijing 100853, China; 6. Tianjin Institute of Meteorological Science, Tianjin 300074, China; 7. The First Clinical Medical College of Lanzhou University ,Lanzhou 730000 ,China)

机构地区：[1]天津市气象台,天津300074 [2]兰州大学大气科学学院甘肃省干旱气候变化与减灾重点实验室,甘肃兰州730000 [3]北京市气候中心,北京100089 [4]成都信息工程大学大气科学学院,四川成都610103 [5]中国人民解放军总医院,北京100853 [6]天津市气象科学研究所,天津300074 [7]兰州大学第一临床医学院,甘肃兰州730000

出　　处：《气象与环境学报》2018年第4期126-133,共8页Journal of Meteorology and Environment

基　　金：天津市气象局重点项目(201723zdxm04);天津市气象局一般项目(201816ybxm10)共同资助

摘　　要：基于2008—2012年北京市4家三甲医院的脑卒中疾病急诊就诊资料及同期气象观测资料和环境监测资料,筛选气象和环境预报因子,采用广义相加(Generalized Additive Model,GAM)、逐步回归、BP(Back Propagation)神经网络及决策树4种方法编辑数据训练集(2008—2011年)和验证集(2012年)输入模型,建立北京市脑卒中疾病预报模型,计算各模型的拟合优度和预报准确率,对比分析脑卒中疾病各预报模型并确定最优预报方法。结果表明:北京市四季脑卒中疾病不同模型选取的预报因子不同,其中时间序列为重要的预报因子。GAM模型对高等级脑卒中疾病的预报效果最好,逐步回归模型对中间等级脑卒中疾病的预报效果最好,决策树模型对低等级脑卒中疾病的预报效果最好。4种脑卒中疾病预报模型四季平均的预报准确率依次为:GAM>神经网络模型>逐步回归模型>决策树。GAM模型脑卒中疾病的平均和高等级预报准确率均为最高,其中出血性脑卒中预报模型的完全预报准确率为68.3%,预报误差≤±1级的准确率达98.0%,可以满足天气变化对出血性脑卒中疾病预报预警的业务需求。Based on the data of stroke emergency visits from four hospitals in Beijing during 2008 to 2012,as well as the observed daily meteorological and environmental factors,meteorological and environmental predictors were selected. All the data were divided into two groups, that is,calibration set（ 2008-2011） and validation set（ 2012）.Causing-stroke weather forecasting model was constructed using four methods including Stepwise Regression Model（ SRM）,BP Neural Network Model,Decision Tree Model（ DTM） and Generalized Additive Model（ GAM）.The best model for forecasting the number of stroke patients was determined by comparing the goodness of fit and forecasting accuracy of different models. The results show that the selected predictors vary along with different seasons and models. The time series is the most important factor. GAM produces the best performance in forecasting number of the high-level-stoke. SRM gives the best result for the prediction of the medium-level-stroke. DTM provides the best estimate for the low-level-stroke patients. The sequence of averaged forecasting accuracy over the four seasons from the three different models is as follows,GAM 〉BP Neural Network Model SRM DTM.GAM has the highest forecasting accuracy for the number of averaged and high-level-stroke patients. For the cerebral hemorrhagic stroke（ CHS）, the forecasting accuracy is 68. 3%. For the bias grade≤ ± 1, the forecasting accuracy is 98%. The result indicates that GAM can basically meet the demand for medical meteorological forecasting of CHS.

关 键 词：医疗气象 脑卒中疾病 广义相加模型GAM BP神经网络 决策树 

分 类 号：P49[天文地球—大气科学及气象学]