Clinical impact detection

Given a collection of plain text documents, systems must (1) perform a document classification according to information relevant to high-impact, real-world clinical use cases (multi-label classification) and (2) retrieve the list of NCBI Tax IDs that support the binary classification. 

Systems have to categorize the documents into the following information axes:

• Pets and farm animals in close contact with the patient (important for detecting animal-transmitted diseases such as toxoplasmosis, salmonellosis, cat-scratch disease, etc.).
• Animals causing injuries. Parasites are NOT included.
• Food species. It includes ingested aliments and any other food mentioned in the document. It excludes ingested items that are not food.
• Nosocomial entities: mentions corresponding to nosocomial/health care associated infections.