Auditing complex concepts in overlapping subsets of SNOMED

Area of Science:

• Medical Informatics
• Ontology Engineering
• Clinical Terminology Management

Background:

• Auditing large terminologies such as SNOMED CT presents significant challenges due to resource limitations and the vast number of concepts.
• Automated methods are needed to assist auditors in identifying high-priority concepts for review.
• Existing auditing techniques may not efficiently pinpoint areas with a higher likelihood of errors.

Purpose of the Study:

• To present a novel methodology for auditing large terminologies, specifically SNOMED CT.
• To algorithmically identify concepts that warrant auditor attention based on a defined abstraction network.
• To evaluate the effectiveness of this methodology in improving the efficiency and accuracy of terminology audits.

Main Methods:

• The study utilizes a previously introduced abstraction network, the p-area taxonomy, for a SNOMED CT hierarchy.
• A methodology is presented that algorithmically identifies concepts within overlapping subsets defined by the p-area taxonomy.
• The methodology was applied to SNOMED CT's Specimen hierarchy and compared against a control group using the double bootstrap method.

Main Results:

• Application of the methodology to the SNOMED CT Specimen hierarchy identified specific concept groups for review.
• Comparison against a control sample showed a statistically significant higher proportion of error discoveries in the group identified by the methodology.
• The double bootstrap statistical method confirmed the significance of the improved error detection rate.

Conclusions:

• The presented methodology offers an effective approach to automatically identify concepts for auditing in large terminologies like SNOMED CT.
• This technique significantly enhances the proportion of error discoveries compared to traditional auditing subset selection.
• The use of p-area taxonomy overlaps provides a valuable tool for optimizing the auditing process of complex biomedical terminologies.