Analysis and prediction of incidence and mortality trends of laryngeal cancer in China

  Objective   Through analysis of incidence and mortality data of laryngeal cancer in China from 1990−2019 in the Global Health Data Exchange (GHDX), the current trend of incidence and mortality was described. Kalman ELM-SVR-GM(1,1) fusion model predicts the future 5-year incidence and mortality data of laryngeal cancer.

  Methods   Joinpoint Regression was used to calculate the Average Annual Percent Change (AAPC) of morbidity and mortality by gender and age, and the trend of morbidity and mortality of laryngeal cancer was described by category. Extreme learning machine model, support vector regression model, and Single variable first-order gray model were trained with the data from 1990 to 2015, and the results were fused by Kalman. The prediction effect of the single model and the fusion model was tested with the data from 2016 to 2019. Finally, the fusion model with the highest prediction accuracy was established based on the data from 1990 to 2019 to predict the morbidity and mortality of various types of laryngeal cancer from 2020 to 2024.

  Results   From 1990 to 2019, the morbidity and mortality of laryngeal cancer increased year by year, morbidity increased from 1.19/100000 to 3.19/100000, and the mortality increased from 0.97/100000 to 1.42/100000. The morbidity and mortality of males were higher, with an average annual increase of 3.93% and 1.72%, respectively. The morbidity for men aged 15~59 increased at the fastest rate of 2.53% per year. Among the six sequence predictions, the KalmanELM-SVR-GM(1,1) fusion model has the lowest average MRE, and its prediction data all fall within the 95%CI.

  Conclusion   The morbidity and mortality of laryngeal cancer showed an upward trend, and the morbidity of males aged 15~49 increased rapidly. Compared with the single model, the Kalman ELM-SVR-GM(1,1) fusion model had a significant impact on the incidence of laryngeal cancer. It has high prediction accuracy and mortality prediction, and can provide a reliable prediction method for laryngeal cancer prevention and control.