Validation of Biosensors for Monitoring Feeding Behaviors in Hanwoo Cattle

Detecting feeding behavior in cattle is important but can be cumbersome when done manually. This study aimed to validate two wearable biosensors─RumiWatch noseband (RW) and Nedap SmartTag Neck (ND)-for monitoring feeding behavior in Hanwoo cattle. Specifically, we assessed the agreement between feeding behaviors recorded by RW and ND and those obtained manually through video observation (VO), focusing on eating time (ET) and ruminating time (RT). Three Hanwoo cows (540 ± 24.12 kg) were fitted with both sensors and continuously video-recorded for three consecutive days. Data from both 1 h (n = 216) and 24 h (n = 9) summaries were used to compare RW with VO, while ND data were analyzed only at the 24 h level. The RW showed strong agreement with VO for RT, with concordance correlation coefficients (CCC) of 0.98 (1 h) and 0.95 (24 h). In contrast, RT from ND showed poor agreement with VO (CCC = 0.12 for 24 h). Least squares means for RT (min/24 h) did not differ between VO (460 min) and RW (461 min), but ND significantly underestimated RT (211 min; p < 0.001). Both RW and ND performed poorly in detecting ET (CCC = 0.05 for RW and -0.15 for ND at 24 h). However, when time spent on both eating and other jaw movements was combined in VO, the combined time did not differ significantly from ET reported by RW (441 min vs. 454 min; p > 0.05), with improved agreement (CCC = 0.62). Cow, day, and their interaction had no significant effect on the deviation between VO and RW for RT (p > 0.05). Overall, these results indicate that RW accurately detects RT but tends to overestimate ET, whereas ND significantly underestimates both ET and RT with poor concordance in Hanwoo cows.