AI and linguistics give a new meaning to the concept of semantic interoperability in electronic medical records systems

Interoperability, the exchange of information from a user of one electronic medical record system (EMR) to a user on another system, is of practical importance and is theoretically an unsolved problem.Most modelling has used ontologies and assumed that meaning of texts and information can be preserved.This elusive goal was named semantic interoperability and was never realised in practice.This work shows, using deep arguments from linguistics, that fundamentally this is a problem stemming from what "meaning" in human communication is.Certain ambiguities are studied in linguistics and cannot be avoided.The problems are especially pertinent if the sender and receiver come from a different culture, have a different medical background, or the receiver is a patient with limited medical knowledge.Also discussed is to what degree large language models can improve on these fundamental linguistic problems and on ontologies.

Hali tarjima qilinmagan