Multi-frequency bioelectrical impedance and machine learning for non-invasive detection of Caseous Lymphadenitis in goats: a one health precision surveillance approach

Caseous lymphadenitis is a chronic infectious disease of sheep and goats caused by Corynebacterium pseudotuberculosis , with major implications for animal health, farm productivity, carcass value, and occupational exposure risk. Because infected animals may remain clinically inapparent or develop chronic lesions, practical field-based screening tools are needed to improve early detection and herd-level disease management. This study evaluated whether multi-frequency bioelectrical impedance analysis could differentiate healthy/control goats from CL-positive goats using resistance, reactance, phase angle, and derived spectral features. Measurements were summarized across 50, 100, and 180 kHz, followed by multivariate visualization, Mahalanobis distance analysis, leave-one-subject-out validation, animal-level risk ranking, calibration assessment, and decision curve analysis. CL-positive goats showed distinct frequency-dependent impedance behavior, including altered resistance-reactance profiles and consistently higher phase angle responses across frequencies. Multivariate spectral distance from the healthy centroid was significantly greater in CL-positive animals, suggesting measurable bioelectrical deviation from the healthy impedance profile. A nonlinear model outperformed the linear model in leave-one-subject-out ROC analysis, with an AUC of 0.68 compared with 0.56, indicating moderate discrimination from raw impedance measurements. Animal-level risk ranking further showed enrichment of CL-positive goats among higher predicted-risk animals, while decision curve analysis suggested potential screening value at lower clinical threshold probabilities. These findings support the feasibility of BIA as a rapid, non-invasive adjunct screening tool for CL-associated risk classification in goats. From a One Health perspective, this approach may improve animal welfare, reduce disease persistence in small-ruminant systems, support producer decision-making, and decrease human exposure risk linked to infected animals and contaminated farm environments.