Reading Cues and Feeding Babies with Behavorial Coding

Magnus Fricker, who began life as a micro-preemie and whose survival was largely dependent on his ability to nurse and feed. Photos by Alexis Kovacevic, above illustration by Eileen Odanaka Vavra.

INTERDISCIPLINARY RESEARCH TEAM'S ANALYSIS OF INFANT CUES MAY LEAD TO HEALTHIER BABIES

APRIL 2023
BY NICK WILSON

The age-old question of how to accurately read and respond to a baby’s feeding cues to support healthy development and growth is the guiding question of a public health and statistics study involving a team of 22 students and two Cal Poly faculty. 

Public health Professor Alison Ventura and statistics Professor Kevin Ross have teamed up to lead a National Institutes of Health-funded research project involving five years of data collected from mothers and infants in a University of Michigan study.

They’re using behavioral coding to compile statistical information from video records of hundreds of mother-infant feeding interactions. Behavioral coding involves quantifying the behaviors babies exhibit to express their needs during feeding and mothers' responses to these behaviors.

“Behavioral coding has been a mainstay of child development research, but it hasn’t been as well studied in the feeding realm,” Ventura said. “What we’re doing is generating much more detail than has ever been seen in the field and answering some novel questions with it.”

The University of Michigan study followed mothers and their babies between the ages of two weeks and 12 months. Ross said that the team tells the story of trends in the data visually with graphs and charts.

“We have a lot of time-dependent data within a feeding, including how a feeding develops over time and how the child develops over different ages,” Ross said. “It’s basically a way of translating what we’re seeing visually to numbers.”

Some of the baby behaviors include negative facial expressions, drooling, falling asleep, crying and pushing the bottle away. The mother’s responses include verbal acknowledgment of hunger or fullness, allowing an infant to hold the bottle, burping the infant and pushing the nipple into infant’s mouth, among many other cues.

The researchers hope their findings will guide educational and clinical efforts to help parents learn and appropriately respond to their baby's cues during feeding, which is thought to be important for supporting healthy growth trajectories and lowering risk for later obesity.

The baby and mother behaviors are coded by a multi-disciplinary team of public health, biology, child development and statistics students.

From left: Kevin James Ross, William Medwid and Alison Ventura make up part of the Cal Poly research team.

“Mostly, what we’ve done is to provide descriptions of the variability and the ways that babies and mothers behave,” Ross said. “In the future, we hope to determine if babies offer cues we haven't yet documented and if we can collect data on other types of interactions between mother and babies and other types of feedings.”

Given the complexity of the data, the statistics students have programmed analytical models for biostatistical analysis of mother-infant interaction during feeding.

Student researcher William Medwid, a statistics senior, said that graphs and models help guide initial open-ended exploration of the data, which has been an essential first step for identifying meaning patterns and trends. They’ve used coding in the R language, a program used widely in statistics and many other fields.

“The coded data includes the exact time of behaviors, and the duration of elongated behavior,” Medwid said. “Most of my statistical analysis was displayed in plots, and some was also displayed in tables for more scientific precision.”

Medwid said R was able to accommodate the research well, but there were a couple of small project-wide systems that he set up.

“I updated the base color palette and standardized the ways behaviors were displayed,” Medwid said.

Ventura said that the team is at the tail end of its analysis and scholarly articles are planned for pediatrics, child development and statistics publications.

“We’ve been working on this project for five years now,” Ventura said. “We’ve had students who have been watching and analyzing these videos for years.”

A survey sample of 325 dyads (mother and infant pairs) is the focus of the research, but because of challenges in the consistency of participants (who may drop out for various reasons) and ability to assess baby behaviors over time, the actual survey size is closer to 150 to 200.

“We’re dependent on what the University of Michigan study provides, but we’re getting really good quality data out of it,” Ross said. “More data is not always better if it’s not very well measured or if it’s not collected in a way that's representative of a broader population.”

Ventura said that an ongoing focus of her career in research at Cal Poly has been to assess parent-child interactions to better understand how to support new parents to promote healthy feeding and development for their infants.

“I’ve always been interested in how parents might be shaping their children, but also how children shape their parents,” Ventura said. “We’re interested in understanding how much babies vary because it's often assumed that all babies are good at communicating their needs, and parents just need to understand what their babies are trying to tell them.”

Ventura said the guiding question is whether babies communicate clearly.

“Maybe some do and some don’t,” Ventura said. “And, if not, we’re interested in understanding whether we can predict which babies are less clear in their ability to tell their moms that they’re full as a first step toward understanding how to give these mothers and infants the targeted support they may need during feeding.”

Medwid added the research “improved my career opportunities by giving me practice working with clients, making useful visualizations and communicating scientific findings. It was likely a beneficial factor in getting my current position at an orthopedic design firm.”

 

 

Graph provided by research group displaying the meal duration in minutes and infant age for each of the dyads included in the research group's analysis.