The embodiment of the invention provides a blood transfusion prediction method, system and device in the liver transplantation process and a medium, and belongs to the technical field of medical care information. The method specifically comprises the steps: training an XGBOOST algorithm through a sample data set, and establishing a prediction model; collecting a key information data set of a target person, wherein the key information data set comprises age, portal hypertension, activated partial thrombin time, direct bilirubin content, globulin content, hemoglobin content, glutamic oxalacetic transaminase content, glutamic-pyruvic transaminase content and liver diagnosis information; and inputting the key information data set into the prediction model to obtain a predicted blood transfusion scheme. According to the scheme, the machine learning algorithm is used for learning the sample data set, the prediction model is established, then the collected key information data set of the target person is input into the prediction model, the predicted blood transfusion scheme is obtained, prediction efficiency and precision are improved, and the safety in the operation process is improved.