本文選題：疑似預(yù)防接種異常反應(yīng) + 馬爾可夫模型　； 參考：《中國(guó)疫苗和免疫》2017年04期

【摘要】：目的應(yīng)用馬爾科夫模型對(duì)甘肅省2016年11月和12月疑似預(yù)防接種異常反應(yīng)(Adverse events following immunization,AEFI)報(bào)告數(shù)進(jìn)行預(yù)測(cè)。方法選取2015年1月-2016年10月甘肅省分月AEFI報(bào)告數(shù),通過(guò)10折交叉驗(yàn)證將其劃分為6個(gè)狀態(tài),通過(guò)時(shí)間與狀態(tài)的轉(zhuǎn)移概率矩陣預(yù)測(cè)2016年11月和12月AEFI報(bào)告數(shù)。結(jié)果通過(guò)轉(zhuǎn)移概率矩陣得到甘肅省2016年11月和12月轉(zhuǎn)移概率分別為(0.33,0.33,0.33,0.00,0.00)和(0.00,0.33,0.19,0.19,0.19,0.08),11月和12月預(yù)測(cè)數(shù)分別為663例和717例,預(yù)測(cè)誤差分別為14.67%和-38.68%。結(jié)論馬爾科夫模型進(jìn)行AEFI報(bào)告趨勢(shì)預(yù)測(cè)是可行的,需要收集較長(zhǎng)的時(shí)間序列數(shù)據(jù)以提高預(yù)測(cè)精度。
[Abstract]:Objective to predict the number of events events following AEFI reports in Gansu Province in November and December 2016 by using Markov model.Methods monthly AEFI reports were selected from January 2015 to October 2016 in Gansu Province, and were divided into 6 states by 10% cross-validation. The number of AEFI reports in November and December 2016 was predicted by the transition probability matrix of time and state.Conclusion it is feasible for Markov model to predict the trend of AEFI report, and it is necessary to collect long time series data to improve the prediction accuracy.
【作者單位】： 甘肅省疾病預(yù)防控制中心;
【分類號(hào)】：O211.62;R186
，

本文編號(hào)：1735027