Abstract: OBJECTIVE To explore the application of different time series models in predicting the detection of carbapenem-resistant Acinetobacter baumannii (CRAB). METHODS Based on the detection rate data of CRAB from the Second Affiliated Hospital of Shantou University Medical College from Jan. 2021 to Sept. 2024, Prophet, SARIMA and Holt-Winters models were established. The data from Oct. 2024 to Mar. 2025 were used for validation to evaluate the prediction effect of the models. RESULTS From 2021 to 2024, the overall detection rate of CRAB showed an upward trend (χ²=4.883, P=0.027). The peak detection period occurred annually from Mar. to Apr., exhibiting periodicity and seasonality. All three models effectively fitted the trend of CRAB detection rates. The Prophet model had a narrower confidence interval at the same confidence level, while the SARIMA model demonstrated the best performance in prediction effect evaluation parameters: root mean square error (RMSE)=13.029, mean absolute percentage error (MAPE)=20.162 and mean absolute error (MAE)=8.932. CONCLUSION All three models exhibit good prediction capabilities for CRAB detection trends and can be utilized for early warning and dynamic analysis of CRAB detection, providing a theoretical basis for the timely formulation and implementation of prevention and control interventions.