Transforming data from one model to another, and in particular to a standard data model, requires deep expertise and domain knowledge. In medicine, where tolerance for error is very low, data transformation is painstakingly slow, expensive and error prone.

That's why we have developed Auto-Mapper, which is included in every data ingestion workflow in Piano. The Auto-Mapper codes text concepts (such as diagnoses, allergies, procedure names and much more) to a standard vocabulary. It uses machine learning and natural language processing to automate quickly and precisely. With it, transforming data from any source to the OHDSI Common Data Model is about 50 times faster than using open source tools.

None of this is worthwhile doing if the precision of the automatic task is not also better than what can be achieved manually. Our recent whitepaper on the precision of the tool is now available here.