Development of a pig wean-quality score using machine-learning algorithms to characterize and classify groups with high mortality risk under field conditions

Aug 5, 2026

Mortality during the post-weaning phase is a critical indicator of swine production performance, shaped by multiple epidemiological factors. Using retrospective data from 1,723 pig groups marketed in a US production system, the study builds a Wean-Quality Score (WQS) via machine learning. Three models are evaluated—Random Forest, Support Vector Machine, and Gradient Boosting Machine—for classifying high versus low 60-day mortality groups. Random Forest achieves the highest performance (accuracy 0.90, sensitivity 0.84, specificity 0.92) and identifies key predictors including pre-weaning mortality, weaning age, sow farm average parity, and PRRS status, plus stocking density and time to fill the barn. The WQS correlates with actual 60-day mortality (r=0.74) and supports targeted producer decisions before weaning.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages35

Reading Time1 h 28 min

AccessFree to download

Abstract

Keywords

Introduction

Post-weaning mortality as a sustainability indicator

Factors and the need for a comprehensive characterization

Importance of weaned pig quality and pre-weaning risk factors

Version of Record: [https://www.sciencedirect.com/science/article/pii/S0167587724002137](https://www.sciencedirect.com/science/article/pii/S0167587724002137)[ ](https://www.sciencedirect.com/science/article/pii/S0167587724002137)[Manuscript_174f44873707df4837933d34736668d6](Manuscript_174f44873707df4837933d34736668d6) Development of a pig wean-quality score using machine-learning algorithms to characterize and classify groups with high mortality risk under field conditions Edison S. Magalhães 1, Danyang Zhang2, Cesar A. A. Moura3, Giovani Trevisan 1, Derald J. Holtkamp 1, Will A. López4, Chong Wang1, 2, Daniel C. L. Linhares 1, Gustavo S. Silva 1 * 1 Department of Veterinary Diagnostic and Production Animal Medicine, College of Veterinary Medicine, Iowa State University, Ames, Iowa; 2 Department of Statistics, College of Liberal Arts and Sciences, Iowa State University, Ames, Iowa; 3 Iowa Select Farms, Iowa Falls, Iowa 4 Pig Improvement Company (PIC), Hendersonville, TN, USA. *Corresponding author Gustavo S. Silva 2221 Lloyd, 1809 S Riverside Drive Department of Veterinary Diagnostic and Production Animal Medicine Iowa State University, Ames, Iowa 50011 [gustavos@iastate.edu](gustavos@iastate.edu) (GSS) © 2024 published by Elsevier. This manuscript is made available under the Elsevier user license [https://www.elsevier.com/open-access/userlicense/1.0/](https://www.elsevier.com/open-access/userlicense/1.0/) 1 1 Development of a pig wean-quality score using machine-learning 2 algorithms to characterize and classify groups with high mortality 3 risk under field conditions 4 5 Author names and information were deidentified as per PVM instructions 6 2 7 Abstract 8 Mortality during the post-weaning phase is a critical indicator of swine production system 9 performance, influenced by a complex interaction of multiple factors of the epidemiological 10 triad. This study leveraged retrospective data from 1,723 groups of pigs marketed within a US 11 swine production system to develop a Wean-Quality Score (WQS) using machine learning 12 techniques. The study evaluated three machine learning models, Random Forest, Support Vector 13 Machine, and Gradient Boosting Machine, to classify groups having high or low 60-day 14 mortality, where high mortality groups represented 25% of the groups among the study 15 population with the highest mortality values (n=431; 60-day mortality=9.98%), and the 16 remaining 75% of the groups were of low mortality (n=1,292; 60-day mortality=2.75%) . The 17 best-performing model, Random Forest (RF), outperformed the other ML models in terms of 18 accuracy (0.90), sensitivity (0.84), and specificity (0.92) metrics, and was then selected for 19 further analysis, which consisted of creating the WQS and ranking the most important factors for 20 classifying groups as high or low mortality. The most important factors ranked through the RF 21 model to classify groups as high with high mortality were pre-weaning mortality, weaning age, 22 average parity of litters in sow farms, and PRRS status. Additionally, stocking conditions such as 23 stocking density and time to fill the barn were important predictors of high mortality. The WQS 24 was developed and correlated (r = 0.74) with the actual 60-day mortality of the groups, offering a 25 valuable tool for assessing post-weaning survivability in swine production systems before 26 weaning. This study highlights the potential of machine learning and comprehensive data 27 utilization to improve the assessment and management of weaned pig quality in commercial 28 swine production, which producers can utilize to identify and intervene in groups, according to 29 the WQS. 3 30 Keywords: swine; wean-quality; 60-day mortality; machine-learning; classification. 4 31 Introduction 32 The swine industry is pivotal in the ever-growing global demand for high-quality protein 33 (Mote and Rothschi

FAQs

What is the Wean-Quality Score (WQS) and what problem does it address?

The WQS is a machine-learning-derived score designed to assess post-weaning survivability. It helps characterize and classify pig groups by their risk of high 60-day mortality under field conditions.

Which machine learning models were tested, and how did Random Forest perform?

Random Forest, Support Vector Machine, and Gradient Boosting Machine were evaluated for classifying high versus low 60-day mortality groups. Random Forest performed best with accuracy 0.90, sensitivity 0.84, and specificity 0.92.

Which factors were identified as most important for predicting high mortality groups?

The Random Forest model ranked pre-weaning mortality, weaning age, average parity of litters in sow farms, and PRRS status as most important. Stocking conditions such as stocking density and time to fill the barn also predicted high mortality.

Aug 5, 2026

Upload by Seraphina