Statistical Analysis and Prediction on Incidence of Infectious Diseases Based on Trend and Seasonality

概要

論文の詳細を見る
We proposed a prediction methodology for the incidence of infectious diseases using incidence data on measles and influenza for forty years in Japan. We also proposed a diagram that makes it possible to convey information on infectious disease incidence more attractively to a wider audience. This can be a useful tool for health promotion in the community.The obtained results are as follows:1. It was advantageous to use data transformed by logarithm in statistical analysis of infectious disease incidence.2. The incidences of measles and influenza exhibited strong seasonality. Measles was most frequent in June and influenza in February.3. Long-term trends were extracted from the derived data obtained by eliminating seasonal effects from the original data. For measles, a decline was accelerated by the introduction of vaccination program in 1978. Influenza also showed a decline for these thirty years.4. The observed incidence data were quite well predicted by only the trend and the seasonality. The squares of multiple regression coefficients of measles and influenza were 0.84 and 0.58, respectively. The analysis of the residuals suggested there was a possibility of improvement in prediction.5. The improvement in prediction was attained by incorporating an autoregressive component of the residuals. As a result, the squares of multiple correlation coefficients of measles and influenza increased to 0.97 and to 0.79, respectively.6. We finally proposed the TS-decomposition diagram to facilitate practical use of incidence data. In this diagram, current incidence data and predicted values for the near future are plotted on the plane where the trend and the seasonality are superimposed. We also discussed the application of our method to the entire range of infectious disease surveillance data.
日本衛生学会の論文

Statistical Analysis and Prediction on Incidence of Infectious Diseases Based on Trend and Seasonality

スポンサーリンク

概要

著者

関連論文

スポンサーリンク