In the field of medical text processing, the task of relation extraction holds significant importance for several reasons. Firstly, it plays a crucial role in the construction of medical knowledge graphs. Secondly, it provides assistance in clinical decision-making processes. Lastly, it contributes to enabling intelligent diagnosis capabilities. Taking into account the intricate nature as well as the high level of professionalism present in Chinese medical data, this paper puts forth a relation extraction model that is built upon the Long Short-Term Memory (LSTM) network. Initially, taking into account the particularities of medical texts, approaches like data cleaning, word segmentation, as well as word vector representation are employed for the pre-processing stage, with the aim of improving the quality of the data. Afterwards, by making use of the sequential modeling capability of LSTM, an architecture for relation extraction that is appropriate for medical texts is crafted. Furthermore, the attention mechanism is incorporated so as to refine the capacity for extracting crucial information. The experimental results clearly show that this model attains a rather high level of accuracy in the task of extracting relations from Chinese medical data. In terms of performance, it even surpasses those traditional methods. At last, this paper thoroughly analyzes the limitations existing in the model and also delves into the potential directions for future improvements.