Cow-level evaluation of plasma essential amino acid clusters and their association with lactating dairy cow performance

Authors: MENEZES, GUIHERME - University Of Wisconsin; LETELIER, PAULINA - University Of Wisconsin; DOREA, JOAO - University Of Wisconsin; Sullivan, Michael; Reinhardt, Laurie; Zanton, Geoffrey; WATTIAUX, MICHAEL - University Of Wisconsin

Submitted to: Journal of Dairy Science
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 1/3/2026
Publication Date: 2/26/2026
Citation: Menezes, G.L., Letelier, P., Dorea, J.R., Sullivan, M.L., Reinhardt, L.A., Zanton, G.I., Wattiaux, M.A. 2026. Cow-level evaluation of plasma essential amino acid clusters and their association with lactating dairy cow performance. Journal of Dairy Science. https://doi.org/10.3168/jds.2025-27455.
DOI: https://doi.org/10.3168/jds.2025-27455

Interpretive Summary: This study demonstrates that supervised machine learning can reliably classify lactating dairy cows into plasma essential amino acid (EAA) profile clusters identified in a previous meta-analysis. A random forest model trained on group-level literature data accurately assigned individual cows to clusters based on plasma EAA concentrations and profiles. These clusters were associated with differences in dry matter intake, milk yield, and nitrogen use efficiency, validating the relationship between EAA profiles and performance at the individual level. Despite higher milk output, cows in the high-EAA cluster showed reduced nitrogen use efficiency. The model’s implementation as a web-based tool supports precision protein nutrition strategies.

Technical Abstract: Studying changes in plasma essential amino acid (EAA) concentrations ([EAA]p) or profile may provide a foundation for precision protein nutrition strategies for feeding dairy cows. In a systematic review, Letelier et al. (2022a) identified two EAA profile clusters that were associated with cow performance, but controlled trials are necessary to evaluate these findings on individual cow-level data. We hypothesized that the amino acid (AA) profile at the group level from the literature would show a similar association with performance and nitrogen use efficiency (NUE) when evaluated at the individual level. Hence, this study aimed to evaluate the clusters identified in the review by Letelier et al. (2022a) using a supervised machine learning model and to evaluate whether the differences in performance and indicators of nitrogen utilization reported in the literature are consistent when assessed for individual cows at four stages of lactation when fed diets containing four levels of crude protein (CP). A random forest algorithm was trained for cluster identification using data from the systematic review, with Arg, His, Ile, Leu, Lys, Met, Phe, Thr and Val concentration as well with the EAA profile as features and clusters as outputs. A leave-one-trial-out cross-validation approach evaluated the model's performance, achieving accuracy, precision, recall, and F1-score of 84.4%, 87.3%, 85.7%, and 86.5%, respectively. The model was then retrained using the entire dataset to predict clusters for 62 cows across 4 lactation stages (e.g., Early, Mid-early, Mid-late, and Late) receiving diets with CP levels of 13.6%, 15.2%, 16.7%, and 18.3% over an 8-week trial (Letelier et al., 2022b). The profiles of EAA were assessed in weeks 4 and 8, where 33 and 38 cows were classified as cluster 1 and 29 and 24 as cluster 2, respectively. Cows in cluster 2 exhibited higher concentrations of Arg (11.8%), His (24.5%), Ile (18.4%), Leu (38.8%), Lys (16%), Met (14.4%), Phe (15.2%), Trp (4.5%), and Val (26.3%) as expected. Cluster 2 cows were also determined to have 2.5% higher (P =0.001) dry matter intake (28.4 ± 0.23 vs. 27.7 ± 0.22 kg/d), produce 2.3% more (P = 0.047) energy corrected milk (42.8 ± 0.51 vs. 43.8 ± 0.54 kg/d), with a 2.8% reduction (P =0.033) in NUE (28.5 ± 0.42 vs. 27.7 ± 0.44 %). Despite increased milk production, the EAA profile was not associated with milk component concentration or feed efficiency. These findings validate the group-level association observed between the AA profile and dairy cow performance at the individual cow level.