Construction and Evaluation of Maternal Mortality Prediction Model in Yunnan Province
-
摘要:
目的 构建并评估云南省孕产妇死亡率预测模型,预测2024—2030年云南省孕产妇死亡率。 方法 基于1994—2023年云南省孕产妇死亡率,构建灰色预测模型和差分自回归移动平均模型,选择平均绝对误差、均方误差和均方根误差比较两种模型回代拟合效果,使用最优模型预测2024—2030年云南省孕产妇死亡率。 结果 1994—2023年云南省孕产妇死亡率整体呈持续下降趋势(χ2 = 50170.0 ,P < 0.05),构建的灰色预测模型和差分自回归移动平均模型平均绝对误差、均方误差和均方根误差分别为2.424、12.389、3.519和3.966、27.651、5.258,灰色预测模型的预测效果优于差分自回归移动平均模型,后验差比值C = 0.079,小概率误差P = 1,预测精确度为1级。用灰色预测模型预测2024—2030年云南省孕产妇死亡率分别为10.05/10万、9.16/10万、8.34/10万、7.59/10万、6.91/10万、6.30/10万、5.73/10万。结论 灰色预测模型对云南省孕产妇死亡率有较好预测效果。经预测,云南省2030年孕产妇死亡率能达到《健康中国“2030”规划纲要》《中国妇女发展纲要(2021—2030年)》《云南妇女发展规划(2021—2030年)》中的孕产妇死亡率控制目标。 -
关键词:
- 孕产妇死亡率 /
- 灰色预测模型 /
- 差分自回归移动平均模型 /
- 预测
Abstract:Objective To construct and evaluate the prediction model of maternal mortality in Yunnan Province, and predict the maternal mortality rate in Yunnan Province from 2024 to 2030. Methods Based on the maternal mortality rates in Yunnan Province from 1994 to 2023, a grey prediction model and a autoregressive integrated moving average model were constructed, The models were compared using mean absolute error, mean square error and root mean square error to assess their fitting performance, and the optimal model was used to predict the maternal mortality rate in Yunnan Province from 2024 to 2030. Resuls The maternal mortality rate in Yunnan Province showed a continuous decline from 1994 to 2023(χ2 = 50170.0 , P < 0.05). The mean absolute error, mean-square error and root mean-square error for the grey prediction model were 2.424, 12.389, 3.519 , respectively, while for the differential autoregressive moving average model, they were 3.966, 27.651, 5.258, respectively. The prediction effect of the grey prediction model is superior to that of the autoregressive integrated moving average model, with a posterior difference ratio C = 0.079 and a low probability error P = 1, indicating a prediction accuracy of level 1. Using the grey prediction model, the maternal mortality rates for Yunnan Province from 2024 to 2030 are 10.05/100 000, 9.16/100 000, 8.34/100 000, 7.59/100 000, 691/100 000, 6.30/100 000 and 5.73/100 000, respectively.Conclusion The grey prediction model has a good prediction effect on maternal mortality in Yunnan Province. It is predicted that the maternal mortality rate in Yunnan Province in 2030 can meet the control targets outlined in the “Healthy China 2030 Plan”, the “Outline of Chinese Women's Development (2021—2030)” and the “Yunnan Women's Development Plan (2021—2030).” -
表 1 两种模型预测结果Tbl.1 Prediction results of the two models
年份 GM(1,1)
偏差率(%)平均偏
差值(%)ARIMA
(1,2,1)
偏差率(%)平均偏
差值(%)1994 / 4.15 / 6.01 1995 −2.14 / 1996 −0.84 −12.81 1997 8.19 −0.45 1998 1.93 −9.79 1999 −2.94 −9.39 2000 0.30 6.38 2001 −8.37 −3.99 2002 −5.18 4.73 2003 −2.47 8.62 2004 0.03 3.59 2005 3.82 3.18 2006 9.84 4.71 2007 1.05 −11.31 2008 4.47 0.13 2009 0.61 −0.26 2010 0.00 −0.53 2011 0.77 3.16 2012 −2.90 −2.29 2013 −1.41 1.97 2014 −3.49 0.11 2015 0.29 4.02 2016 2.01 2.71 2017 0.29 −3.5 2018 0.27 −1.57 2019 −1.58 −1.58 2020 −2.20 −0.45 2021 −1.41 1.8 2022 0.15 2.12 2023 −1.34 −1.93 表 2 两种模型的指标数据值比较
Table 2. Comparison of indicator data values between the two models
模型 指标数据值比较 MAE MSE RMSE GM(1,1) 2.4238 12.39 3.52 ARIMA 3.9659 27.65 5.25 MAE,平均绝对误差;MSE,均方误差;RMSE,均方根误差。 -
[1] 尹小兰,何芯芯,杜林,等. RNN、JPR及ARIMA 3种模型预测中国农村孕产妇死亡率的比较[J]. 中华疾病控制杂志,2023,27(11):1308-1313. [2] Organization W H . Health in 2015: From MDGs to SDGs[J]. 2015. [3] 国务院. 国务院关于印发中国妇女发展纲要和中国儿童发展纲要的通知[EB/OL]. (2021-09-08)[2024-06-26] -https://www.gov.cn/gongbao/content/2021/content_5643262.htm [4] 云南省人民政府. 《云南妇女发展规划(2021—2030年)》[EB/OL](2022-05-31)[2024-06-26] https://www.ynzx.gov.cn/gwfb/264869.jhtml [5] 王嫚,杨伟康. 基于ARIMA模型与GM(1,1)模型的深圳市5岁以下儿童死亡率的预测研究[J]. 现代预防医学,2024,51(3):466-470+506. [6] 刘洁,高茵茵,曲波,等. 应用ARIMA模型预测我国孕产妇死亡率[J]. 中国医科大学学报,2011,40(02):107-108+121. [7] 庞艳蕾. 孕产妇、婴儿及5岁以下儿童死亡率分析与预测[D]. 潍坊: 潍坊医学院,2016. [8] 邵升清,夏桂梅. ARIMA模型与GM(1,1)模型在传染病发病率中的预测效果比较[J]. 宁夏师范学院学报,2021,42(7):13-18. [9] Yan J M,Li Y B,Zhou P Y. Impact of COVID-19 pandemicon the epidemiology of STDs in China: Based on the GM(1,1) model[J]. BMC Infect Dis,2022,22(1):519. doi: 10.1186/s12879-022-07496-y [10] Zeng B,Yang Y J,Gou X. Research on physical health early warning based on GM(1,1)[J]. Comput Biol Med,2022,143(1):105256. [11] Cheng T,Bai Y,Sun X,et al. Epidemiological analysis of varicella in Dalian from 2009 to 2019 and application of three kinds of model in prediction prevalence of varicella[J]. BMC Public Health,2022,22(1):678. doi: 10.1186/s12889-022-12898-3 [12] 张悦,郭海强. 1991—2017年我国妇幼保健主要统计指标趋势分析[J]. 中国卫生统计,2019,36(5):642-643. [13] 联合国可持续发展峰会. 变革我们的世界: 2030年可持续发展议程: [EB/OL] (2015-09-27) [2024-06-28] https://www.un.org/zh/conferences/environment/newyork2015 [14] 世界卫生组织. 2000年至2020年孕产妇死亡率趋势: 世界卫生组织、儿童基金会、人口基金、世界银行集团和经社部/人口司的估计数[EB/OL] : (2023-02-23) [2024-6-28]. https://www.who.int/publications/i/item/9789240068759 [15] 刘昶荣. 我国妇幼健康核心指标达到历史最好水平[N]. 中国青年报,2022-08-23(004). [16] Samuel,O,Zewotir,T,North,D. Decomposing the urban-rural inequalities in the utilisation of maternal health care services: Evidence from 27 selected countries in Sub-Saharan Africa. Reprod Health. 2021; 18 (1): 216. [17] 陈敦金,贺芳. 中国孕产妇死亡率极大程度降低——对世界的最大贡献[J]. 中国实用妇科与产科杂志,2019,35(10):1076-1080. [18] 苏燕燕,神艳,聂雷,等. 1991—2021年我国孕产妇死亡率变化情况分析[J]. 中国初级卫生保健,2024,38(1):54-58. doi: 10.3969/j.issn.1001-568X.2024.01.0015 [19] 王震坤,王子伟,沈霞,等. 1992—2017年中国孕产妇死亡趋势及年龄—时期—队列模型分析[J]. 中华预防医学杂志,2021,55(6):742-746. doi: 10.3760/cma.j.cn112150-20210127-00083 [20] 黄翠敏,吴桂花,粘惠瑜,等. 海南省2003—2022年孕产妇死亡率变化趋势及预测分析[J]. 中国热带学,2023,23(11):1164-1168+,1179. [21] 谭晓萍. 云南省实现联合国可持续发展目标中的妇幼保健和生殖健康相关目标研究[D]. 昆明: 昆明医科大学,2019. [22] 张亚慧,张荣,贾素红,等. 灰色预测模型GM(1,1)和残差自回归模型在中国孕产妇死亡率预测中的应用[J]. 中国妇幼保健,2023,38(10):1747-1752. [23] 张彬,张龙秀,王瑞,等. 基于ARIMA和GM(1,1)模型的我国农村MMR、IMR、NMR及U5MR趋势预测[J]. 中国医药科学,2023,13(14):7-10+15.