Abstract
Human brucellosis remains a significant public health issue in the Ili Kazak Autonomous Prefecture, Xinjiang, China. To assist local Centers for Disease Control and Prevention (CDC) in promptly formulate effective prevention and control measures, this study leveraged time-series data on brucellosis cases from February 2010 to September 2023 in Ili Kazak Autonomous Prefecture. Three distinct predictive modeling techniques-Seasonal Autoregressive Integrated Moving Average (SARIMA), eXtreme Gradient Boosting (XGBoost), and Long Short-Term Memory (LSTM) networks-were employed for long-term forecasting. Further, the optimal model will be used to explore the impact of COVID-19 on the transmission of Human brucellosis in the region. We constructed a SARIMA(4,1,1)(3,1,2)12 model, an XGBoost model with a time lag of 22, and an LSTM model featuring 3 LSTM layers and 100 neurons in the fully connected layer to predict monthly reported cases from January 2021 to September 2023. The results indicated that the occurrence of brucellosis exhibits pronounced seasonal patterns, with higher incidence during summer and autumn, peaking in June annually. Performance evaluations revealed low Mean Absolute Error (MAE), Root Mean Square Error (RMSE), and Symmetric Mean Absolute Percentage Error (SMAPE) for all three models. Specifically, the coefficient of determination (R(2)) was 0.6177 for the SARIMA model, 0.8033 for the XGBoost model, and 0.6523 for the LSTM model. The study found that the XGBoost model outperformed the other two in long-term forecasting of brucellosis, demonstrating higher predictive accuracy. This discovery can aid public health departments in advancing the deployment of prevention and control resources, particularly during peak seasons of brucellosis. It was also found that the impact of the COVID-19 pandemic on the transmission of human brucellosis in the region was minimal. This research not only provides a reliable predictive tool but also offers a scientific basis for formulating early prevention and control strategies, potentially reducing the spread of this disease.