时间序列和机器学习模型在上海市中小学生因病缺课趋势预测中的应用

王正中; 张喆; 周欣怡; 袁琳琳; 翟娅妮; 孙力菁; 罗春燕

doi:10.16835/j.cnki.1000-9817.2025082

时间序列和机器学习模型在上海市中小学生因病缺课趋势预测中的应用

doi: 10.16835/j.cnki.1000-9817.2025082

上海市疾病预防控制中心儿童青少年健康所, 200336

基金项目:

国家社会科学基金项目 22BTY105

上海市加强公共卫生体系建设三年行动计划(2023—2025年)重点学科项目 GWVI-11.1-31

上海市加强公共卫生体系建设三年行动计划(2023—2025年)学科带头人项目 GWVI-11.2-XD14

详细信息

作者简介:
王正中(1997-)，女，浙江杭州人，硕士，主要从事学校卫生工作

通讯作者:
孙力菁，E-mail：sunlijing@scdc.sh.cn

罗春燕，E-mail：luochunyan@scdc.sh.cn

利益冲突声明 所有作者声明无利益冲突。
中图分类号: R179 N945.12 G478
计量
- 文章访问数: 58
- HTML全文浏览量: 32
- PDF下载量: 6
- 被引次数: 0
出版历程
- 收稿日期: 2024-04-02
- 修回日期: 2024-12-04
- 网络出版日期: 2025-04-03
- 刊出日期: 2025-03-25

Application of time series and machine learning models in predicting the trend of sickness absenteeism among primary and secondary school students in Shanghai

Institute of Child and Adolescent Health, Shanghai Municipal Center for Disease Control and Prevention, Shanghai (200336), China

摘要

摘要: 目的分析上海市中小学生因病缺课的时间变化规律，探索适用于预测因病缺课高峰时间和强度的模型。方法运用季节趋势分解(STL)法分析上海市中小学生2010年9月1日至2018年6月30日因病缺课率的季节性和长期趋势变化，并采用基于动态时间规整(DTW)的分层聚类方法对时间变化模式相似的缺课症状进行分类。基于历史数据，构建并评估不同时间序列算法和机器学习模型，以优化预测因病缺课趋势的准确性。结果研究期间学年平均新发因病缺课率为16.86/万人天，缺课趋势呈现季节性和长期上升特征，在2017学年达到最高(22.47/万人天)。缺课症状可分为3类：冬春高发类(呼吸道症状、发热及全身不适等)、夏季高发类(眼部症状、鼻出血等)以及无明显季节性类(皮肤症状、意外伤害等)。构建的时间序列模型均能较好地预测因病缺课时间趋势，但峰值强度预测的准确度较低，其中多层感知器模型表现最好，均方根误差为8.96，平均绝对误差为4.37，相较基准模型分别降低36.51%和39.02%。结论时间序列模型和机器学习算法能够有效预测因病缺课的趋势，在高峰期可针对不同症状导致的缺课采取相应防控措施。
- 时间 /
- 序列分析 /
- 疾病 /
- 学生
Abstract: Objective To analyze the temporal variation patterns of sickness absenteeism among primary and secondary school students in Shanghai, so as to explore models suitable for predicting peaks and intensity of absenteeism rates. Methods The seasonal and trend decomposition using loess (STL) method was used to analyze the seasonal and long-term trend changes in sickness absenteeism among primary and secondary school students from September 1 in 2010 to June 30 in 2018, in Shanghai. A hierarchical clustering method based on Dynamic Time Warping (DTW) was employed to classify absenteeism symptoms with similar temporal patterns. Based on historical data, the study constructed and evaluated different time series algorithms and machine learning models to optimize the accuracy of predicting the trend of sickness absenteeism. Results During the research period, the average new absenteeism rate due to illness was 16.86 per 10 000 person-day for every academic year, and the trend of sickness absenteeism exhibited both seasonality and a long-term upward trend, reaching its highest point in the 2017 academic year (22.47 per 10 000 person-day). The symptoms of absenteeism were divided into three categories: high incidence in winter and spring (respiratory symptoms, fever and general discomfort, etc.), high incidence in summer (eye symptoms, nosebleeds, etc.) and those without obvious seasonality (skin symptoms, accidental injuries, etc.).The constructed time series models effectively predicted the trend of absenteeism due to illness, although the accuracy of predicting peak intensity was relatively low. Among them, the multi-layer perceptron (MLP) model performed the best, with an root mean squared error (RMSE) of 8.96 and an mean absolute error (MAE) of 4.37, reducing 36.51% and 39.02% compared to the baseline model. Conclusion Time series models and machine learning algorithms could effectively predict the trend of sickness absenteeism, and corresponding prevention and control measures can be taken for absenteeism caused by different symptoms during peak periods.
- Time /
- Sequence analysis /
- Disease /
- Students
注释:

1) 利益冲突声明 所有作者声明无利益冲突。

HTML全文

图 1 上海市中小学生新发因病缺课率时间序列季节趋势分解

Figure 1. Seasonal and trend decomposition of the time series of sickness absenteeism rate among primary and secondary students in Shanghai

下载: 全尺寸图片幻灯片

图 2 上海市中小学生不同聚类因病缺课症状季节变化趋势

Figure 2. Seasonal trends in symptoms of sickness absenteeism in different clusters among primary and secondary students in Shanghai

下载: 全尺寸图片幻灯片

图 3 不同模型预测2017—2018学年中小学生新发因病缺课情况

Figure 3. Prediction of sickness absenteeism in 2017-2018 based on different models among primary and secondary students

下载: 全尺寸图片幻灯片

表 1 不同模型误差指标与改善率的比较

Table 1. Comparison of errors metrics and improvement rates of different models

模型	RMSE	MAE
SNAIVE	14.11(0.00)	7.16(0.00)
SARIMA	12.92(8.40)	5.91(17.54)
TBATS	12.10(14.26)	6.65(7.18)
SVR	11.40(19.18)	5.84(18.52)
MLP	8.96(36.51)	4.37(39.02)
注： ()内数字为改善率/%。

下载: 导出CSV

参考文献(25)

[1]	马军. 中国学校卫生/儿少卫生发展[J]. 中国学校卫生, 2015, 36(1): 6-9. http://www.cjsh.org.cn/article/id/zgxxws201501002 MA J. School health/ child and adolescent health development in China[J]. Chin J Sch Health, 2015, 36 (1): 6-9. (in Chinese) http://www.cjsh.org.cn/article/id/zgxxws201501002
[2]	TSANG T K, HUANG X, GUO Y, et al. Monitoring school absenteeism for influenza-like illness surveillance: systematic review and Meta-analysis[J]. JMIR Public Health Surveill, 2023, 9: e41329. doi: 10.2196/41329
[3]	DONALDSON A, HARDSTAFF J, HARRIS J, et al. School-based surveillance of acute infectious disease in children: a systematic review[J]. BMC Infect Dis, 2021, 21(1): 744. doi: 10.1186/s12879-021-06444-6
[4]	SCHMIDT W P, PEBODY R, MANGTANI P. School absence data for influenza surveillance: a pilot study in the United Kingdom[J]. Eur Surveill, 2010, 15(3): 19467.
[5]	郝莉, 朱冰, 施文英, 等. 2013—2017学年浙江省杭州市中小学生因病缺课监测分析[J]. 疾病监测, 2020, 35(9): 840-844. HAO L, ZHU B, SHI W Y, et al. Surveillance for illness-induced school absence in students of elementary and secondary schools in Hangzhou, 2013-2017[J]. Dis Surveill, 2020, 35(9): 840-844. (in Chinese)
[6]	BOX G E P, JENKINS G M, REINSEL G C, et al. Time series analysis: forecasting and control[M]. Hoboken, New Jersey: John Wiley & Sons Inc, 2016: 88-106.
[7]	HYNDMAN R J, KHANDAKAR Y. Automatic time series forecasting: the forecast package for R[J]. J Stat Softw, 2008, 27(3): 1-22.
[8]	DE LIVERA A M, HYNDMAN R J, SNYDER R D. Forecasting time series with complex seasonal patterns using exponential smoothing[J]. J Am Stat Assoc, 2011, 106(496): 1513-1527. doi: 10.1198/jasa.2011.tm09771
[9]	SMOLA A J, SCHÖLKOPF B. A tutorial on support vector regression[J]. Stat Comput, 2004, 14(3): 199-222. doi: 10.1023/B:STCO.0000035301.49549.88
[10]	RUMELHART D E, HINTON G E, WILLIAMS R J. Learning representations by back-propagating errors[J]. Nature, 1986, 323(6088): 533-536. doi: 10.1038/323533a0
[11]	SVETUNKOV I, PETROPOULOS F. Old dog, new tricks: a modelling view of simple moving averages[J]. Int J Prod Res, 2018, 56(18): 6034-6047. doi: 10.1080/00207543.2017.1380326
[12]	KARA E O, ELLIOT A J, BAGNALL H, et al. Absenteeism in schools during the 2009 influenza A(H1N1) pandemic: a useful tool for early detection of influenza activity in the community?[J]. Epidemiol Infect, 2012, 140(7): 1328-1336. doi: 10.1017/S0950268811002093
[13]	胡停停, 赵鹤鹤, 段晓健, 等. 2013—2020年我国急性出血性结膜炎流行特征及暴发疫情分析[J]. 疾病监测, 2021, 36(5): 440-444. HU T T, ZHAO H H, DUAN X J, et al. Epidemiological characteristics and outbreak analysis of acute hemorrhagic conjunctivitis in China, 2013-2020[J]. Dis Surveill, 2021, 36(5): 440-444. (in Chinese)
[14]	LUTZ C S, HUYNH M P, SCHROEDER M, et al. Applying infectious disease forecasting to public health: a path forward using influenza forecasting examples[J]. BMC Public Health, 2019, 19(1): 1659. doi: 10.1186/s12889-019-7966-8
[15]	孙婕, 杨雯雯, 曾令佳, 等. 2011—2016年全国6~22岁学生人群法定传染病监测数据分析[J]. 中华流行病学杂志, 2018, 39(12): 1589-1595. doi: 10.3760/cma.j.issn.0254-6450.2018.12.010 SUN J, YANG W W, ZENG L J, et al. Surveillance data on notifiable infectious diseases among students aged 6-22 years in China, 2011-2016[J]. Chin J Epidemiol, 2018, 39 (12): 1589-1595. (in Chinese) doi: 10.3760/cma.j.issn.0254-6450.2018.12.010
[16]	黄硕, 刘才兄, 邓源, 等. 世界主要国家和地区传染病监测预警实践进展[J]. 中华流行病学杂志, 2022, 43(4): 591-597. doi: 10.3760/cma.j.cn112338-20211105-00856 HUANG S, LIU C X, DENG Y, et al. Progress in the practice of surveillance and early warning of infectious diseases in major countries and regions[J]. Chin J Epidemiol, 2022, 43 (4): 591-597. (in Chinese) doi: 10.3760/cma.j.cn112338-20211105-00856
[17]	CHOWELL G, SATTENSPIEL L, BANSAL S, et al. Mathematical models to characterize early epidemic growth: a review[J]. Phys Life Rev, 2016, 18: 66-97.
[18]	沈忠周, 马帅, 曲翌敏, 等. ARIMA模型在我国法定传染病报告数中的应用[J]. 中华流行病学杂志, 2017, 38(12): 1708-1712. doi: 10.3760/cma.j.issn.0254-6450.2017.12.025 SHEN Z Z, MA S, QU Y M, et al. Application of autoregressive integrated moving average model in predicting the reported notifiable communicable diseases in China[J]. Chin J Epidemiol, 2017, 38 (12): 1708-1712. (in Chinese) doi: 10.3760/cma.j.issn.0254-6450.2017.12.025
[19]	PETROPOULOS F, APILETTI D, ASSIMAKOPOULOS V, et al. Forecasting: theory and practice[J]. Int J Forecast, 2022, 38(3): 705-871. doi: 10.1016/j.ijforecast.2021.11.001
[20]	付之鸥, 周扬, 陈诚, 等. 时间序列分析与机器学习方法在预测肺结核发病趋势中的应用[J]. 中国卫生统计, 2020, 37(2): 190-195. FU Z O, ZHOU Y, CHEN C, et al. Application of time series analysis and machine learning methods in predicting the incidence of tuberculosis[J]. Chin J Health Stat, 2020, 37 (2): 190-195. (in Chinese)
[21]	杨慧, 魏麟, 胡晓斌, 等. 基于CEEMD-GWO-SVR集成模型的病毒性肝炎流行趋势预测研究[J]. 中国卫生统计, 2022, 39(6): 815-818. YANG H, WEI L, HU X B, et al. Epidemiological trend prediction of viral hepatitis based on CEEMD-GWO-SVR integration method[J]. Chin J Health Stat, 2022, 39 (6): 815-818. (in Chinese)
[22]	孔德川, 吴寰宇, 郑雅旭, 等. 上海市2015—2017年成年人急性呼吸道感染病例的流行病学和病原学特征分析[J]. 中华流行病学杂志, 2019, 40(8): 904-910. doi: 10.3760/cma.j.issn.0254-6450.2019.08.007 KONG D C, WU H Y, ZHENG Y X, et al. Etiologic and epidemiologic features of acute respiratory infections in adults from Shanghai, during 2015-2017[J]. Chin J Epidemiol, 2019, 40 (8): 904-910. (in Chinese) doi: 10.3760/cma.j.issn.0254-6450.2019.08.007
[23]	YAO Y, CHEN L, ZHU D, et al. Increasing serum antibodies against type B influenza virus in 2017-2018 winter in Beijing, China[J]. AMB Express, 2022, 12(1): 127.
[24]	张晓波, 施鹏, 郑珊, 等. 上海市单中心儿科门诊特征和医疗服务2009—2018年趋势分析[J]. 中国循证儿科杂志, 2019, 14(3): 161-168. ZHANG X B, SHI P, ZHENG S, et al. An analysis of characteristics and trends of outpatient visits and medical services in a tertiary pediatric hospital in Shanghai from 2009 to 2018[J]. Chin J Evid-Based Pediatr, 2019, 14(3): 161-168. (in Chinese)
[25]	张欣. 新发传染病是学校卫生面临的永恒挑战[J]. 中国学校卫生, 2020, 41(6): 641-644. ZHANG X. Perpetual challenges of emerging infectious disease of school children[J]. Chin J Sch Health, 2020, 41 (6): 641-644. (in Chinese)