Sensitivity and calibration disparities of a race-blind machine learning model for periodontitis across racial and ethnic groups

Maria Fernanda Escalante Freitez and Mafefreitez@gmail.com

SAP Artificial Intelligence in Dentistry, 2026

Abstract: Introduction: Machine learning models are increasingly proposed for periodontitis prediction, but their fairness across racial and ethnic groups is rarely quantified. Methods: A race-blind logistic regression model was trained on 12,630 adults aged 30 years and older from NHANES 2009-2014 to predict periodontitis, and its discrimination, sensitivity, specificity, calibration, and demographic parity were evaluated within each racial and ethnic group. Results: The model discriminated periodontitis with an overall area under the curve of 0.656. Sensitivity ranged from 54.1% to 79.2% and specificity from 46.0% to 73.9% across groups, yielding an equal-opportunity gap of 25.1 percentage points and a demographic-parity gap of 29.5 points, while calibration error remained below 12.2% in every group. Conclusions: A race-blind periodontitis model produced marked disparities in sensitivity and specificity across racial and ethnic groups through proxy variables, despite comparable calibration.

Date: 2026
References: Add references at CitEc
Citations:

Downloads: (external link)
https://southam.pub/journals/files/aid/aid2026142en.pdf (application/pdf)

Related works:
This item may be available elsewhere in EconPapers: Search for items with the same title.

Export reference: BibTeX RIS (EndNote, ProCite, RefMan) HTML/Text

Persistent link: https://EconPapers.repec.org/RePEc:cwf:aidart:aid2026142

DOI: 10.62486/aid2026142

Access Statistics for this article

More articles in SAP Artificial Intelligence in Dentistry from South American Publishing
Bibliographic data for series maintained by South American Publishing Journals Manager ().