Validation of Dairy Cow Bodyweight Prediction Using Traits Easily Recorded by Dairy Herd Improvement Organizations and Its Potential Improvement Using Feature Selection Algorithms.

Anthony Tedde,Isabelle Dufrasne,Frédéric Dehareng,Zhiquan Wang,Nicolas Gengler,Clément Grelet,Gpluse Consortium Group ,Mark Crowe,Phuong Ho,Hélène Soyeurt,Yves Brostaux,Jennie Pryce,Graham Plastow,Dagnachew Hailemariam,Carlo Bertozzi,Eric Froidmont

doi:10.3390/ani11051288

Abstract

Simple SummaryFirst, the current work consisted of validating the feasibility of large-scale dairy cow bodyweight prediction from models involving the day in milk, milk yield, parity, and milk mid-infrared spectrum. Second, it aimed to improve the accuracy of predictive models by using feature selection algorithms to decrease the number of predictors to limit overfitting. The models, using accessible and low-cost measurements, provided highly reproducible predictions. These could be easily obtained on an individual basis throughout a cow’s productive life by dairy herd improvement organizations, thus providing potentially relevant information for the dairy farmer at three levels: economics (reproductive performance), animal welfare (disease detection), and environment (methane production).Knowing the body weight (BW) of a cow at a specific moment or measuring its changes through time is of interest for management purposes. The current work aimed to validate the feasibility of predicting BW using the day in milk, parity, milk yield, and milk mid-infrared (MIR) spectrum from a multiple-country dataset and reduce the number of predictors to limit the risk of over-fitting and potentially improve its accuracy. The BW modeling procedure involved feature selections and herd-independent validation in identifying the most interesting subsets of predictors and then external validation of the models. From 1849 records collected in 9 herds from 360 Holstein cows, the best performing models achieved a root mean square error (RMSE) for the herd-independent validation between 52 ± 2.34 kg to 56 ± 3.16 kg, including from 5 to 62 predictors. Among these models, three performed remarkably well in external validation using an independent dataset (N = 4067), resulting in RMSE ranging from 52 to 56 kg. The results suggest that multiple optimal BW predictive models coexist due to the high correlations between adjacent spectral points.

Highlights

Knowing the live bodyweight (BW) of a dairy cow is an important source of information related to farm management [1], disease detection [2], quantification of dry matter intake [3], estimation of methane production [4,5], or even assessing the reproductive performance of a given cow through its body weight (BW) changes [6]
The of the analysis focused on the three models with the best RMSE of validation (RMSEv) sta-9 of 15 tistics (52 and 56 kg): HSO_VIP, HSO_BETA, and HSO_SBF_BETA
Multiple optimal models coexisted in predicting BW using MIR spectra, it was shown that the usage of these spectra within a modeling process to predict BW is relevant, and the performance achieved were comparable with such offered by methods using linear morphological traits as input variables

Summary

Introduction

Knowing the live bodyweight (BW) of a dairy cow is an important source of information related to farm management [1], disease detection [2], quantification of dry matter intake [3], estimation of methane production [4,5], or even assessing the reproductive performance of a given cow through its BW changes [6]. Even if BW is measured at farms, the data often stay unshared, limiting its use by Dairy Herd Improvement (DHI) organizations. Previous studies showed the value of using observations from linear morphological classification in isolation [9,10], or in combination [11,12,13] in predicting BW, with accuracy ranging from 36 to 105 kg, and R2 varying between 0.53 and 0.99. Accurate repeated measurements of morphological traits would be required, which could be costly and labor-intensive. Linear morphological classification estimations are seldom taken during an animal’s lifetime [8,14], which would prevent controlling BW over time

Objectives

Methods

Results

Conclusion