A Combined Model of SARIMA and Prophet Models in Forecasting AIDS Incidence in Henan Province, China

Abstract

Acquired immune deficiency syndrome (AIDS) is a serious public health problem. This study aims to establish a combined model of seasonal autoregressive integrated moving average (SARIMA) and Prophet models based on an L1-norm to predict the incidence of AIDS in Henan province, China. The monthly incidences of AIDS in Henan province from 2012 to 2020 were obtained from the Health Commission of Henan Province. A SARIMA model, a Prophet model, and two combined models were adopted to fit the monthly incidence of AIDS using the data from January 2012 to December 2019. The data from January 2020 to December 2020 was used to verify. The mean square error (MSE), mean absolute error (MAE), and mean absolute percentage error (MAPE) were used to compare the prediction effect among the models. The results showed that the monthly incidence fluctuated from 0.05 to 0.50 per 100,000 individuals, and the monthly incidence of AIDS had a certain periodicity in Henan province. In addition, the prediction effect of the Prophet model was better than SARIMA model, the combined model was better than the single models, and the combined model based on the L1-norm had the best effect values (MSE = 0.0056, MAE = 0.0553, MAPE = 43.5337). This indicated that, compared with the L2-norm, the L1-norm improved the prediction accuracy of the combined model. The combined model of SARIMA and Prophet based on the L1-norm is a suitable method to predict the incidence of AIDS in Henan. Our findings can provide theoretical evidence for the government to formulate policies regarding AIDS prevention.

Spatial‐temporal dynamics and time series prediction of HFRS in mainland China: A long‐term retrospective study

“…The Prophet prediction model shows that 14 223 cases of HFRS are expected to occur in mainland China from 2019 to 2038. The Prophet model has been applied within the environment, COVID‐19 and AIDS, and the results have shown better efficacy 44,45 . Compared with traditional models such as ARIMA, Holt‐Winters model, hybrid support vector regression, neural network autoregressive model, and so forth, the prophet model uses machine learning algorithms and is also able to predict time series data accurately 46,47 .…”

“…and AIDS, and the results have shown better efficacy. 44,45 Compared with traditional models such as ARIMA, Holt-Winters model, hybrid support vector regression, neural network autoregressive model, and so forth, the prophet model uses machine learning algorithms and is also able to predict time series data accurately. 46,47 Thus, it can be estimated that the improved performance of the prophet model in this study has a better prediction effect of HFRS and can be used for predicting infectious diseases such as HFRS in the future.…”

Time series models in prediction of severe fever with thrombocytopenia syndrome cases in Shandong province, China

“…There may be discrepancies between the forecasted data and the actual data in certain months. Some studies have shown that combining linear and nonlinear models may yield better predictive performance than using a single model, such as SARIMA-Prophet( Luo et al., 2022 ), Prophet-LSTM ( Na et al., 2019 ), SARIMA-NARX ( Wang et al., 2019 ), etc. Therefore, further research is needed to enhance the predictive ability.…”

Prediction research of AIDS monthly incidence cases and mortality cases based on prophet model

“…h(t) represents the impact of an irregular schedule that may occur on holiday or more days. ε is the error term and assumed the normal distribution in the Luo study [8].…”