Objective To establish an early warning model of healthcare-associated infection(HAI) in the department of neurology, predict the risk of HAI in patients in department of neurology, and provide basis for early prevention and control. Methods Data on incidence of HAI in neurology ward of a tertiary first-class hospital in Guizhou Province were collected, the ARIMA(p,d,q)×(P,D,Q)s model was constructed, parameter estimation and model diagnosis were performed for the established model, and the optimal prediction model was selected. The best constructed model was used to predict the incidence of HAI in the department of neurology, and the prediction efficacy was evaluated. Results The data of monthly incidence of HAI in department of neurology in this hospital from 2014 to 2017 was as training specimens, the optimal prediction model ARIMA(2,1,2)×(1,1,1)₄ was obtained. Data of January-May 2018 was as validation sample for model prediction, the results showed that the dyna-mic trend of predicted value of model was basically consistent with the actual condition, the actual incidence was within the 95% confidence interval of the predicted value. This model was used to predict the incidence of HAI in department of neurology from June to December 2018, the predicted results showed that the predicted values were within 95% confidence interval. Conclusion The ARIMA(2,1,2)×(1,1,1)₄ model can better simulate the trend of HAI rate in the department of neurology, and it has preferable prediction effect.