Newborn piglets face a particular challenge—their mother, or sow, often has fewer teats than the number of piglets in the litter.

In the U.S., up to one in four piglets die before finishing nursing, a high preweaning mortality rate driven by multiple factors. This not only presents economic challenges for producers but takes an emotional toll on caretakers and raises animal welfare concerns.

Benjamin and a cross-disciplinary, multi-institution research team are using artificial intelligence (AI) to better understand piglet feeding patterns. Their goal is to develop solutions to reduce preweaning mortality, improve production efficiency, and ensure the welfare of all piglets.

The National Institute of Food and Agriculture (NIFA) has awarded Benjamin and the team a $1 million grantover four years for their proposal, IDEAS Tripartite: Automated Piglet and Sow Monitoring for Early Detection of At-Risk Piglets. This project is one of 11 selected for the Inter-Disciplinary Engagement in Animal Systems (IDEAS) program, which supports integrated research and outreach projects focused on precision animal management, the environmental impacts of animal production, and the societal aspects of animal welfare.

The primary goal is to generate robust data to inform sow and piglet husbandry practices and contribute to research in areas such as nutrition science, breeding, lactation biology, and animal sciences, including genetics and genomics.

A Tripartite Team, Cascading Knowledge

This international team, or tripartite, is made up of global leaders from North America, the Republic of Ireland, and Northern Ireland, specializing in computer vision, data-driven technologies, sustainable animal production, global food security, production animal welfare, and piglet and sow nutrition and welfare. With research sites in three countries, the team will collect data from farms using diverse husbandry methods, including the free-farrowing pens used in Ireland.

“Our goal is to capture a ground truth on piglet nursing that will inform not only our own research but also provide a resource for researchers in other disciplines,” says Benjamin. “Through the engineering expertise of Drs. Daniel Morris, Niall McLaughlin, and Tami Brown-Brandl, our computer vision capture systems will see nuances that human eyes can’t—each camera will focus on one sow and her piglets 24 hours a day without the distraction of other tasks.”

Sow Health, Nursing Patterns, and Teat Fidelity

Sows and piglets display complex nursing and suckling behaviors that significantly impact piglet health. Piglets typically select one teat and remain loyal to it throughout nursing, a behavior known as teat fidelity. The team aims to develop a method to automatically track piglets' nursing patterns and teat preferences, as well as monitor sow behavior. They will use AI to engineer an automated system to track piglet movement and nursing activity.

This project will also provide data to support research into piglet selection and teat fidelity. “We will use computer vision to identify the size of the sow’s teats early in the lactation process,” says Benjamin. “This is particularly important because the number of piglets often exceeds the number of available teats.”

Producers typically assess the number of teats when the sow gives birth, but earlier research from collaborators, including Dr. Chantal Farmer, Dr. Russel Hovey, Laura Boyle, Dr. Ilias Kyriazakis, Dr. Edgar Manzanilla, and Dr. Ramon Muns, shows that piglet survival and growth rely on early access to colostrum and milk. Non-functional teats, however, may not be visually apparent during the first three days of lactation. If the project can successfully quantify teat function as proposed by Dr. Tami Brown-Brandl, producers could be alerted early to non-functional teats, which could directly improve production practices.

“Willingness to Play”: Data, Productivity, and Animal Husbandry

Currently, there is little consistency in standard operating procedures across U.S. farms, reducing the likelihood of widespread improvements in piglet vitality. Data is crucial for inspiring large-scale changes in practices.

“Producers stay in business when they’re profitable,” says Benjamin. “They make operational decisions based on data. Sometimes, data reveals that practices believed to increase profitability — like moving piglets from nursing pens too early — actually reduce productivity and profits.”

Collaborators Dr. Mark Knauer, Dr. Eduardo Beltranena, and Dr. Beth Ferry are engaging producers and other stakeholders to gain insights into production practices, research methods, and strategies for generating buy-in from the farming community and other stakeholders.

In the final year, the team will develop training materials with caretakers featuring animated characters, or avatars. Dr. Monique Pairis-Garcia will lead surveys to assess caretaker opinions on the use of AI and its impact on job satisfaction, a concept termed “Willingness to Play.”

“We’re really excited about this project,” says Benjamin. “NIFA IDEAS reviewers called the proposal ambitious.”

PigLetArt: A Capstone Exhibition

An interactive exhibition, PigLetArt, will cap off the project. This exhibit will feature video reels and pixelated 2D images of piglet behavior collected during the research. It will be showcased at four international campuses: Michigan State University, North Carolina State University, the University of Nebraska-Lincoln, and Queen's University Belfast. Visitors will have the opportunity to engage with virtual education modules that allow them to "save piglets."

The IDEAS Tripartite Team

The IDEAS Tripartite: Automated Piglet and Sow Monitoring for Early Detection of At-Risk Piglets team includes: Drs. Ilias Kyriazakis and Niall McLaughlin from Queen’s University Belfast; Dr. Ramon Muns from the Agri-Food and Biosciences Institute in Belfast; Drs. Edgar Garcia Manzanilla and Laura Boyle from the Agriculture and Food Development Authority of the Republic of Ireland; Drs. Monique Pairis-Garcia, Mark Knauer, and Eduardo Beltranena from North Carolina State University; Dr. Tami Brown-Brandl from the University of Nebraska-Lincoln; Dr. Russ Hovey from the University of California, Davis; Dr. Chantal Farmer from the Sherbrooke Research and Development Centre of Agriculture and Agri-Food Canada; and Drs. Beth Ferry, Daniel Morris, and Madonna Benjamin from Michigan State University.

This work is supported by grant no 2024-68014-42559 from the USDA National Institute of Food and Agriculture, Ireland’s Department of Agriculture, Food and the Marine, and the Northern Ireland Department of Agriculture, Environment, and Rural Affairs as part of the US-Ireland Research and Development Partnership.