Development and validation of a rule-based algorithm to identify periodontal diagnosis using structured electronic...

Area of Science:

• Periodontology
• Dental Informatics
• Artificial Intelligence in Healthcare

Background:

• Accurate periodontal diagnosis relies on the 2017 World Workshop classification.
• Electronic Health Records (EHR) contain valuable clinical data for diagnosis.
• Automating diagnostic suggestions can support clinical decision-making.

Purpose of the Study:

• To develop and validate an EHR-based algorithm for periodontal diagnosis.
• To align diagnostic suggestions with the 2017 World Workshop criteria.
• To assess the algorithm's accuracy in staging and grading periodontal disease.

Main Methods:

• Iterative development of a rule-based algorithm using EHR clinical data (CAL, probing depth, etc.).
• Validation through manual chart reviews by expert periodontists.
• Refinement of the algorithm based on expert feedback and data limitations.

Main Results:

• Initial algorithm accuracy: 71.8% for stage, 64.7% for stage and grade.
• Refined algorithm accuracy: 79.6% for stage, 68.8% for stage and grade.
• Demonstrated moderate accuracy in suggesting periodontal diagnoses.

Conclusions:

• EHR-based algorithms can offer moderate accuracy for periodontal diagnosis support.
• The tool is particularly beneficial for less experienced clinicians.
• Algorithm performance is contingent on EHR data quality and completeness.