中国中医眼科杂志2026，Vol.36Issue(6)：509-516,8.DOI:10.13444/j.cnki.zgzyykzz.2026.06.002
基于ARIMA模型对呼和浩特市新生儿眼病筛查发生趋势及相关因素的分析
Analysis of the Incidence Trend and Related Factors of Neonatal Eye Disease Screen-ing in Hohhot Based on ARIMA Model
摘要
Abstract
OBJECTIVE To analyze the incidence trend of neonatal eye disease screening in Hohhot based on the ARIMA model and explore related factors.METHODS A total of 2,581 neonates(5,162 eyes)born in Hohhot Maternal and Child Health Hospital from November 2020 to October 2022 who participated in eye disease screening were included.Analysis was conducted based on different types of eye diseases and different months.After classification,their distribution was analyzed,and an ARIMA multiplicative seasonal prediction model was established.For high-risk factors of neonatal eye diseases,univariate analysis was performed first,followed by screening using multivariate Logistic regression analysis.A regression association analysis model was applied to analyze the differences between the distribution of neonatal eye disease types and the distribution of high-risk factors in Hohhot.RESULTS(1)Distribution of neonatal eye diseases:There were 564 cases(21.85％)with abnormal screening results.The numbers of abnormal cases in 2020,2021,and 2022 censuses were 42(22.70％),289(22.47％),and 233(20.99％),respectively.(2)Monthly distribution of neonatal eye diseases:No seasonal pattern was observed in the screening months for neonatal eye diseases,and the difference was not statistically significant(P>0.05).(3)Time series analysis:The series exhibited no periodicity or seasonality and did not meet the stationarity requirements for model construction.(4)Series stationarization:After differencing,the model parameters were determined as d=1,D=1,and s=12.(5)Model identification and order determination:The ARIMA(0,1,1)(0,1,1)12 model was the optimal model.(6)Estimation and model diagnostics:The Ljung-Box test for white noise yielded,the difference was not statistically significant(P>0.05).(7)Model prediction performance evaluation:The actual numbers of abnormal eye disease screenings from 2023 to 2024 all fell within the 95％CI limits of the predicted numbers,indicating good predictive performance of the model.(8)Factors influencing abnormal neonatal eye disease screening:Preterm birth,maternal use of anti-infective medicines,birth weight<2.5 kg,birth asphyxia or hypoxia,delivery mode,and intrapartum infection were factors influencing abnormal neonatal eye disease screening,with statistically significant differences(all P<0.05).(9)Multivariate Logistic regression analysis of high-risk factors for abnormal neonatal eye disease screening:Preterm birth,maternal use of anti-infective drugs,birth weight<2.5 kg,birth asphyxia or hypoxia,delivery by eutocia,and intrapartum infection were risk factors for abnormal neonatal eye disease screening,with statistically significant differences(all P<0.05).CONCLUSIONS Abnormal neonatal eye disease screening mainly involves retinal hemorrhage and peripheral leukoplakia.Its occurrence is closely related to perinatal factors such as anti-infective drugs use,delivery mode,and preterm birth,with no seasonal pattern.The ARIMA model can effectively predict its short-term change trend.关键词
ARIMA模型/新生儿眼病/筛查/发生趋势/相关因素Key words
ARIMA model/neonatal eye disease/screening/incidence trend/related factors分类
医药卫生引用本文复制引用
刘俊珍,刘桂云,姚文斌,李洋..基于ARIMA模型对呼和浩特市新生儿眼病筛查发生趋势及相关因素的分析[J].中国中医眼科杂志,2026,36(6):509-516,8.基金项目
呼和浩特市卫生健康领域科技项目(呼卫健医疗-2023065) （呼卫健医疗-2023065）
