Referent tracking: the problem of negative findings

Area of Science:

• Medical Informatics
• Ontology Engineering
• Clinical Data Management

Background:

• Referent tracking in Electronic Health Records (EHRs) relies on realist presuppositions, excluding 'negative entities'.
• Handling 'negative findings' (e.g., absence of a condition) poses a challenge for such systems.
• Existing EHR models may struggle to accurately represent the absence of clinical observations.

Purpose of the Study:

• To investigate how referent tracking-based EHRs can effectively manage negative findings.
• To develop an ontological framework capable of representing negative findings without positing negative entities.

Main Methods:

• Analysis of 748 sentences from patient charts containing negation.
• Development of a novel ontological relationship to represent the absence of specific clinical attributes.

Main Results:

• A new ontological relationship was identified: a particular and a universal, where no instance of the universal has a specific qualified relation to the particular.
• This relation successfully accommodated nearly all negative findings in the analyzed sample.
• The proposed method avoids the need to introduce spurious negative entities into the ontology.

Conclusions:

• Referent tracking-based EHRs can manage negative findings through a novel ontological relation.
• This approach enhances the accurate representation of clinical information by modeling absence effectively.
• The findings support a more robust and realistic ontological foundation for EHR systems.