Artificial Intelligence (AI) has undergone considerable development in recent years in the field of medicine and in particular in decision support diagnostic. However, the development of such algorithms depends on the presence of a sufficiently large amount of data to provide reliable results. Unfortunately in medicine, it is not always possible to provide so much data on all pathologies. This problem is particularly true for rare diseases. In this paper we focus on uveitis, a rare disease in ophthalmology which is the third cause of blindness worldwide. This pathology is difficult to diagnose because of the disparity in prevalence of its etiologies. In order to provide physicians with a diagnostic aid system, it would be necessary to have a representative dataset of epidemiological profiles that have been studied for a long time in this domain. This work proposes a breakthrough in this field by suggesting a methodological framework for the generation of an open source dataset based on the crossing of several epidemiological profiles and using data augmentation techniques. The results of these generated synthetic data have been qualitatively validated by specialist physicians in ophthalmology. Our results are very promising and consist in a first brick to promote research in AI on Uveitis disease.