Rethinking navel size thresholds for omphalitis diagnosis in newborn dairy calves using machine learning

Graphical Abstract Summary: Omphalitis in newborn calves is an overlooked disease with no gold-standard diagnostic test. Umbilical stump diameter has been proposed as a proxy for clinical status, but existing thresholds relied on expert opinion, and their clinical performance is unknown. Using machine learning leveraged by veterinary expertise and observed samples from 437 Holstein and 230 Jersey newborn heifers across 17 dairy farms in California (US), our study proposed new thresholds to diagnose omphalitis. This methodology allows for estimating the clinical performance of the diagnostic test without the need to determine the clinical condition of the study subjects. We propose 2 thresholds: one to estimate farm-level prevalence (16.5 mm for Holstein, 13.0 mm for Jersey) and another to identify calves likely to be diseased (19.3 mm for Holstein, 14.9 mm for Jersey).