The aim of this repository is to build a model that can predict, from the data characterizing each patient, whether he will have complications during the operation. Knowing that the more methods tested, the more it will be possible to find the best algorithm to answer the problem. This allowed us, after testing several techniques, to opt for the ensemble learning method which is based on the construction of a predictive model by pooling other models that are more or less efficient. Thus, we were able to create a graphical interface that uses the trained predictive model and displays the prediction to the user based on the patient number entered. This aims to facilitate and simplify the use of the model built.