From tests to truth: A misclassification-aware machine learning framework for estimating brucellosis seroprevalence...

Area of Science:

• Veterinary Medicine
• Zoonotic Disease Epidemiology
• Wildlife Health

Background:

• Brucellosis is a significant zoonotic disease with under-estimated prevalence in wildlife.
• Limited surveillance and biased sampling in wild species hinder accurate Brucellosis exposure assessment.
• Wild canids are potential reservoirs and disseminators of Brucella species (spp.)

Purpose of the Study:

• To estimate Brucellosis exposure patterns in free-ranging wild canids globally.
• To clarify true seroprevalence by correcting for imperfect diagnostic test performance.
• To identify geographic areas needing improved One Health surveillance for Brucellosis.

Main Methods:

• Systematic literature search of Brucella spp. prevalence studies in wild canids (1962-2025).
• Analysis of serological data using a misclassification-aware, multi-assay model.
• Separate consideration of confirmatory polymerase chain reaction (PCR) and culture results for active infection.

Main Results:

• Global true seroprevalence in wild canids was estimated at 8.2% (95% CI: 5.1-11.3%).
• Confirmed active Brucella infection (PCR/culture) was uncommon, at 3.9% (95% CI: 3.0-5.1%).
• Seroprevalence varied geographically, with higher estimates in South America (~18%) and lower in Europe (~0.8%) and North America (~4.1%).

Conclusions:

• Wild canids show widespread but modest serological exposure to Brucella spp.
• Confirmed active Brucellosis infections are rare in wild canids.
• Improved diagnostic rigor and targeted One Health surveillance are crucial, especially in Africa and Asia.