2018-2022年哈密地区结核分枝杆菌感染趋势及基因型分布特征

Trends and genotype distribution of Mycobacterium tuberculosis in Hami regions from 2018 to 2022

    摘要
  • 摘要:
    目的  采用自回归移动平均(ARIMA)模型构建时间序列，对2018-2022年哈密地区结核分枝杆菌(MTB)感染趋势进行分析，并分析哈密地区MTB基因型分布特征。
    方法  收集2018年1月-2022年12月期间于哈密市中心医院就诊的结核病患者经临床分离培养获得的788株MTB菌株，McSpoligotyping分型后分析2018-2022年哈密地区MTB基因型分布情况；建立ARIMA乘积季节预测模型，确定模型(p, d, q)(P, D, Q)s参数；预测2023年MTB感染数并与实际发生数进行比较，评价模型的预测效果。
    结果  2018-2022年哈密地区788株MTB经McSpoligotyping分型结果显示，北京家族基因型共419株、EAI家族24株、H家族64株、T家族79株、LAM3家族1株、Manu2家族3株、X1家族2株、所属家族未定义的有195株，其中北京家族基因型占比最高(53.17%)；2018-2022年3-5月为哈密地区MTB感染高峰期，其中4月占比最高(13.58%)，建立ARIMA模型，确定最佳模型为ARIMA(0, 1, 1)(0, 1, 1)12，其AIC值和BIC值分别为3.258和3.460，经白噪声检验Ljung-Box Q=18.797差异无统计学意义(P=0.279)；以2023年MTB感染病例为样本，对比实际检出数量与预测数值，结果显示所有实际检出数量均落在预测数值的95%CI内，模型RMSE、MAPE、MAE分别为4.988、22.977和3.646，提示模型的预测效果较好。
    结论  哈密地区MTB感染类型主要以北京家族基因型为主。哈密地区MTB感染具有季节性特征，高峰期集中于每年的3-5月，其中4月份占比最高。ARIMA模型在预测MTB感染的短期流行趋势方面表现出色。

     

    Abstract:
    OBJECTIVE  To analyze the trend of Mycobacterium tuberculosis (MTB) infection using time series analysis of autoregressive moving average (ARIMA) model, and the distribution characteristics of MTB genotypes in Hami regions from 2018 to 2022.
    METHODS  Totally 788 strains of MTB were isolated from tuberculosis patients treated in our hospital from Jan. 2018 to Dec. 2022, and the distribution of MTB genotypes was analyzed by McSpoligotyping. The seasonal ARIMA model and the determined model parameters (p, d, q)(P, D, Q)s predicted the number of MTB infections in 2023. By comparing to the observed infection data in 2023, the prediction effect of the model was evaluated.
    RESULTS  From 2018 to 2022, totally 788 strains of MTB were identified by McSpoligotyping, of which 419 strains belonged to Beijing family with the highest proportion 53.17%, other family strains included EAI (n=24), H (n=64), T (n=79), LAM3 (n=1), Manu2 (n=3), X1(n=2) and other undefined family(n=195). The peak period of MTB infection was during Mar. and May, with a peak in April (13.58%). The best ARIMA model was established to be ARIMA(0, 1, 1)(0, 1, 1)12, of which AIC and BIC values were 3.258 and 3.460, respectively, there was no significant difference by white noise test(Ljung-Box Q=18.797, P=0.279).The actual values observed in 2023 were well within bounds of 95% confidence interval of our selected ARIMA prediction model; and ARIMA model performed good with RMSE, MAPE, MAE of 4.988, 22.977 and 3.646, respectively.
    CONCLUSIONS  The Beijing family genotype is the predominant for MTB infection in Hami region. Seasonality exists and the peak period of MTB infection is from March to May with the highest peak in April annually. ARIMA model has a good performance in predicting the short-term prevalence of MTB infection.

     

    • HTML全文
    • 参考文献(24)
    • 相关文章
    • 施引文献
  • 资源附件(0)

/

返回文章
返回
下载中心