中国甲状腺癌发病趋势分析与预测

目的:分析2003-2017年我国患甲状腺癌的不同群体的发病率情况，采用KELM-SVR耦合模型建模并对2018-2022年甲状腺癌发病率进行预测，为甲状腺癌防治提供有益补充。方法:收集2003-2017年全国总体、男性、女性、城市及农村人口的甲状腺癌发病率，建立KELM、SVR、KELM-SVR耦合模型，以MRE为准则，选择精度最高的KELM-SVR耦合模型对2018-2022年不同甲状腺癌发病率进行预测。结果:KELM-SVR耦合模型在五种不同甲状腺癌发病率的预测中均优于KELM、SVR模型，KELM、SVR、KELM-SVR模型的平均MRE分别为:7.58%、6.59%、5.74%，2018-2022年我国总体甲状腺癌发病率分别为:17.07/10万、18.40/10万、19.80/10万、21.23/10万、22.71/10万。结论:甲状腺癌发病率处于稳定上升趋势，其中女性及城市人口甲状腺癌发病率最高，KELM-SVR耦合模型可提高单模型的预测精度，对预测多种甲状腺癌发病率提供稳定可靠的方法。

Analysis and prediction of the incidence trend of Chinese thyroid cancer

Objective:Through analysis of incidence of thyroid cancer in different groups in my country from 2003 to 2017,the KELM-SVR coupling model was used to model and predict the incidence of thyroid cancer from 2018 to 2022,and provide a useful supplement for the prevention and treatment of thyroid cancer.Methods:Collect the national total,male,female,urban and rural populations' incidence of thyroid cancer from 2003 to 2017.Establish KELM,SVR and KELM-SVR models,and select the KELM-SVR coupling model with the highest accuracy based on MRE to predict the incidence of different thyroid cancers from 2018 to 2022.Results:The KELM-SVR coupled model was superior to the KELM and SVR models in predicting the incidence of five different thyroid cancers.The average MRE of the KELM,SVR,and KELM-SVR models was 7.58%,6.59%,and 5.74%,respectively.The incidence of thyroid disease in my country was 17.07/100 000,18.40/100 000,19.80/100 000,21.23/100 000,and 22.71/100 000 from 2018 to 2022.Conclusion:The incidence of thyroid cancer is on a steady upward trend,and the incidence of thyroid cancer is the highest in women and urban populations.The KELM-SVR coupled model can improve the prediction accuracy of a single model and provide a stable and reliable method for predicting the incidence of various thyroid cancers.