使用R语言对两种结核病发病预测模型的对比分析  

Comparison of Two Predicting Models for Tuberculosis Incidence Using R Language

作　　者：汪洋[1] 胡代玉 杨德香 刘瑛 王润华[4] 易静[4] 

机构地区：[1]重庆医科大学实验教学管理中心公共卫生实验教学中心,重庆市401331 [2]重庆市结核病防治所 [3]重庆市南岸区妇幼保健院 [4]重庆医科大学公共卫生与管理学院卫生统计学教研室

出　　处：《中国全科医学》2013年第13期1498-1501,共4页Chinese General Practice

基　　金：国家自然科学基金(30872160);重庆市科委自然科学基金(CSTC2009BB5415)

摘　　要：目的使用R语言分析模型对结核病发病率数据进行演算与分析,探索较佳的分析模型。方法使用重庆市1993—2007年结核病发病登记数作为建模数据,采用R语言中的神经网络模型和灰色模型分别预测2008年与2009年的发病人数,并与其对应年份的真实发病人数进行比较。结果采用神经网络模型预测结果误差分别为4.75%、0.83%,灰色模型误差分别为27.08%、49.84%。结论神经网络模型为较佳分析模型,预测结果对实际工作的量化和资源调配有重要的指导意义。Objective To compare two predicting models for tuberculosis （TB） incidence using R language. Methods Recorded numbers of TB patients in Chongqing between 1993 and 2007 were used as modeling data. Neuro - network model and gray model programmed with R language were used to predict the TB incidence in 2008 and 2009, respectively. These pre- dicted TB incidences were compared with the real TB incidences in 2008 and 2009. Results The predicting errors of neuro - net- work model were 4.75% for 2008 and 0. 83% for 2009, while those of gray model were 27.08% for 2008 and 49. 84% for 2009. Conclusion The neuro - network model is more accurate in predicting TB incidence than the gray model.

关 键 词：结核病 R语言 预测模型 

分 类 号：R181.8[医药卫生—流行病学]