For new parents struggling with breastfeeding, getting an accurate assessment of their newborn’s nursing ability is critical – but current evaluation methods have major limitations. Now, researchers from the University of California San Diego have developed a novel system that leverages a modified pacifier and AI algorithms to objectively analyze infant suckling patterns.
The study, published in the IEEE Journal of Translational Engineering in Health and Medicine, describes how the simple pacifier device connects to a vacuum sensor that records detailed suckling data as the baby feeds. Machine learning software then analyzes that data against established norms, flagging potential abnormalities in suckling strength or rhythmic patterns.
“It’s reassuring to rely on this scientific data to back up our assessments,” said Erin Walsh, a speech-language pathologist at UC San Diego Health and co-author of the study. “We hope these findings can help support families struggling with breastfeeding and improve long-term health outcomes.”
Currently, clinicians subjectively evaluate breastfeeding through two main methods: monitoring the infant’s weight gain, and feeling suckling intensity by placing a finger in the baby’s mouth – a technique the AI system’s data shows could be improved.
In a proof-of-concept study with 91 healthy newborns, the researchers found the device validated most clinical evaluations but also detected abnormal suckling patterns in five infants that providers had missed. It also revealed how commonly performed tongue-tie surgery did not improve nursing mechanics in half of the babies who received it.
“Our measurement system offers rapid, precise data on an infant’s ability early on, empowering clinicians to get to root causes quickly and potentially reduce breastfeeding difficulties,” explained lead author Phuong Truong.
The researchers aimed to create an affordable, accessible screening tool by using off-the-shelf components like the standard pacifier. With breastfeeding challenges a major factor in many new parents discontinuing nursing, the team hopes the AI-driven insights could prevent unnecessary treatments and interventions.
“Frenotomies are not a one-size-fits all solution for breastfeeding issues,” Walsh said, noting the tenfold increase in the tongue-tie surgery over the past decade. “This data shows we need more personalized, objective evaluation to determine which babies truly require intervention.”
Looking ahead, the researchers plan to conduct further testing and refinement of the AI system, with the goal of bringing an affordable screening product to clinical settings. By leveraging innovative technologies like machine learning, the team aims to provide greater support and improved breastfeeding outcomes for families.
