Comparison of a Laser Methane Detector with the GreenFeed for On-Farm Measurements of Enteric Methane Mmissions From Goats

The current study aimed to evaluate the diagnostic accuracy of the Laser Methane Detector (LMD) compared to the GreenFeed (GF) system for measuring methane emissions in goats. It employed two analytical frameworks to analyze agreement: continuous agreement of paired readings and diagnostic accuracy at a designated clinical limit of 13 g/day, using a significance level of α = 0.05. The study involved 45 Spanish Boer goats aged 4–6 months, with methane levels measured over 14 days. Goats had free access to the GreenFeed unit, and readings were taken using the LMD, directed at the nostrils for one minute at a fixed distance. The results showed that the accuracy was 47.6% (95% CI 39.7–55.5%), the no-information rate was 59.1%, and p (Acc > NIR) = 0.9989, meaning it was not better than guessing the majority class. The sensitivity was 34.0% (missed ~2/3 of GF positives) and the specificity was 67.2% (moderate ability to correctly call negatives). Similarly, the PPV/NPV was 60.0%/41.3%, respectively, with balanced accuracy of 50.6%. In terms of the agreement, Kappa was 0.011—essentially indicating no agreement beyond chance. McNemar’s test indicated that p = 9.82×10⁻⁶, with many more false negatives (64) than false positives (22); this means that the LMD systematically under-calls positives at this threshold. The experiment concluded that the LMDr’s diagnostic performance against GF is insufficient for stand-alone use at the current threshold (AUC≈0.54; sensitivity≈0.34; κ≈0.01). Finally, there was no clear correspondence between GF and LMD device measurements. Given the under-calling of positives and near-chance discrimination, the LMD should not replace GF without recalibration and threshold optimization.