
Each day, the Rogers Hixon Ontario Human Milk Bank pools and pasteurizes milk from three to five carefully chosen donors. This process creates a special “recipe” of nutritious milk for preterm infants.
Proper nutrition at this stage is essential for their growth and brain development.
“For a variety of reasons, many hospitalized infants do not have a full supply of their parent’s milk,” says Dr. Sharon Unger, a neonatologist and medical director of the Milk Bank. “In this instance, donor human milk can be life-saving, particularly as it helps protect preterm infants from necrotizing enterocolitis (NEC), a life-threatening bowel disease.”
Today, many milk banks rely on individual decision-making when they mix donor human milk. Donor milk is very important to critically ill preterm infants in neonatal intensive care units (NICUs). Canada does not yet have a standardized method or approved device to analyze donor milk. Without this, it is difficult to ensure a steady supply with consistent macronutrients.
At the University of Toronto, Professor Timothy Chan External Link and Mechanical and Industrial Engineering PhD student Rachel Wong conducted a multistep study. They developed a new model that uses data and machine learning to find the best ways to mix donor milk for vulnerable infants.
The study included a one-year test at the Milk Bank. The results showed that using machine learning increased fat and protein in pools of donor milk. It also reduced the time needed to create recipes by 60 per cent.
It is important to make sure the milk that preterm and critically ill infants drink is not too high in protein or fat. Machine learning has helped the Milk Bank team refine its recipes. Now, the team can make sure each batch of donor milk has the right amount of protein and calories. The new data-driven optimization model was published in Manufacturing and Service Operations Management.
“I hope that this research will provide a framework to help milk banks across North America increase the consistency of macronutrient content in their donor milk product,” says Dr. Unger. “The eventual end goal would be to see a downstream impact of improved growth and developmental outcomes for the infants who receive this donor milk.”
