J Anim Sci. 2025 Nov 6:skaf394. doi: 10.1093/jas/skaf394. Online ahead of print.
ABSTRACT
Accurate live prediction of carcass and meat quality traits is essential for enhancing product consistency and production efficiency in the lamb industry. Traditional assessments rely on postmortem measurements, which require animal slaughter and thus cannot be used for repeated measurements or on-farm decision making. In this study, we developed a modified bioelectrical impedance analysis (BIA) system incorporating multi-site needle electrodes to collect full-body resistance data from 204 live crossbred lambs. These electrical features, combined with basic body size measurements, were used to construct predictive models for 10 carcass and meat quality traits using linear regression and six machine learning algorithms. Among these, multiple linear regression showed the best overall performance, with R2 values exceeding 0.70 for traits such as carcass weight, abdominal fat, and meat color. Feature importance analysis indicated that resistance values from specific anatomical regions were strongly associated with fat deposition, muscle structure, and water-holding capacity. Our findings demonstrate that this integrated BIA-based approach provides a practical, low-cost, and minimally invasive method for live-animal phenotyping, offering valuable applications in on-farm meat quality screening, precision nutrition, and genetic selection programs.
PMID:41206537 | DOI:10.1093/jas/skaf394
