The project aims to develop a model capable of reliably flagging biomedical articles (appearing on bioRxiv or in biomedical scientific publications) that may be at risk of retraction. Such articles would then be carefully reviewed by peers in the community.