Our AI beingis written in Python by following a deep learning segmentation approach, using CNN (Convoluted Neural Network) architec-turearchitecture. Deep learning is a subcategory of ma-chinemachine learning, in which the AI is given more freedom to find a significant pattern itself. This enables it to learn from the traintrained set of pic-turespictures and their correlated body composition measurements (like Sex, CRP, etc.). Next, with the test set, the AI is trying to predict the now missing correlated body composition meas-urementsmeasurements from pictures (Funduscopy, OCT, OCTA) alone. This kind of method is called Su-pervisedSupervised, as the AI is given labeled measure-mentsmeasurements in the training step. We want to keep our choice for the used model open, as more effective models can be released throughout the project.