Purpose/Significance To predict the monthly consumption of the anti-allergic eye drops in the outpatient pharmacy of hospitals, and to provide references for the refined drug procurement and inventory management in hospitals. Method/Process Taking the monthly consumption of anti-allergic eye drops in the outpatient pharmacy of Beijing Chaoyang Hospital, Capital Medical University from January 2013 to December 2023 as the time series, the SARIMA model is constructed using SPSS 29.0 software, and the SARIMA-LSTM combined model is constructed using Python 3.12.1 software. The predictive effects of the two models are verified based on the data from October 2021 to December 2023, and the two models are applied to make short-term predictions on the consumption of anti-allergic eye drops from January to June 2024. Result/Conclusion The SARIMA-LSTM hybrid model demonstrates lower average absolute error (MAE) , mean absolute percentage error (MAPE) and root mean square error (RMSE) in both the validation dataset and the future prediction compared to the SARIMA model. The SARIMA-LSTM hybrid model provides better forecasting performance for the drug consumption and has practical application value.