Venturus/AnonyMED-BR

AnonyMED-BR是一个为巴西葡萄牙语医疗文本匿名化研究创建的数据集。它结合了真实的电子健康记录（EHRs）和合成的医疗记录，以支持健壮的基于变压器的模型开发。该数据集的语言为巴西葡萄牙语，领域为临床/医疗，任务包括命名实体识别（NER）、去识别化和匿名化。数据集大小为2,962个样本。由于伦理和法律限制，仅公开数据集的合成部分。

AnonyMED-BR is a dataset created for research on medical text anonymization in Brazilian Portuguese. It combines real electronic health records (EHRs) with synthetically generated medical records to support the development of robust transformer-based models. The dataset is in Brazilian Portuguese, in the domain of clinical/medical, with tasks including Named Entity Recognition (NER), De-identification, and Anonymization. The dataset size is 2,962 samples. Due to ethical and legal constraints, only the synthetic portion of the dataset is publicly available.