某医院2010年至2014年超声诊断的六大类致死性畸形发病情况及ARIMA预测模型的建立*
Incidence of six categories of fatal malformation diagnosed by ultrasound detection in a hospital from 2010 to 2014 and establishment of ARIMA prediction model

目的：分析某医院2010年至2014年超声诊断的六大类致死性畸形发病情况，并应用自回归移动平均模型(ARIMA) 建立发病预测模型。方法：收集某医院2010年1月至2014年12月采用四维超声系统筛查发现的六大类致死性畸形病例资料，分析其发病规律；采用2010年1月至2013年12月的数据拟合ARIMA 模型，并用该模型对2014 年1月至12月的发病率进行预测。结果：2010年1月至2014年12月共检查86 018例，筛查出六大类致死性畸形胎儿1 063例，总发病率为12.36‰，其中发病率最高的为严重开放性脊柱裂（3.67‰），其后依次为单腔心（3.39‰）、腹壁裂？材谠嗤夥？（2.06‰）、无脑儿（1.77‰）、致死性骨发育不良（0.79‰）、严重脑膨出（0.67‰）。ARI(1，0)模型能够较好地预测六大类致死性畸形月发病率，具有较高的预测精度（MAD为0.381 7，MAPE为28.01%，MSE为0.061 8）。结论：六大类致死性畸形的发病率仍较高，ARI（1,0）模型能较好地用于其发病率的预测。
To analyze the incidence of six categories of fatal malformation diagnosed by ultrasound detection in a hospital, and apply auto？？regressive integrated moving average(ARIMA) to establish the prediction model. Methods: The data of six categories of fatal malformation diagnosed by ultrasound detection in a hospital from 2010 January to 2014 December were collected and analyzed. ARIMA model was established by fitting the incidence from 2010 January to 2013 December, then was used to predict the incidence from 2014 January to December. The predicted and actual values were compared in order to test the predicting effect. Results: A total of 1 063 cases of six categories of fatal malformation were detected from 86 018 fetuses, the total incidence was 12.36‰, and the highest was severe open spina bifida(3.67‰), followed by single chamber(3.39‰), abdominal fissure？？visceral valgus(2.06‰), anencephaly(1.77‰), thanatophoric dysplasia (0.79‰), and severe encephalocele(0.67‰). ARI (1,0) model could better forecast the incidence of six categories of fatal malformation, and had higher prediction accuracy(MAD=0.381 7，MAPE=28.01%，MSE=0.061 8). Conclusion: The incidence of six categories of fatal malformation is still high, and ARI(1,0) model can better predict it