Significance To study the extraction method of adverse drug event (ADE) data from medical texts, and to provide support for clinical drug risk management and scientific decision-making. Method/Process Based on pre-trained model, by combining the correlation between the two subtasks of entity recognition and relation extraction, a entity relation joint extraction method for ADE monitoring is designed.Result/Conclusion Experiments on the published ADE extraction dataset show that the proposed method is superior to existing methods and can effectively extract ADE information and its relation from medical texts, providing a powerful means for the discovery and monitoring of ADE.