Breast milk from donors can vary a lot in nutrition depending on the mother’s age, diet, and other factors. This is especially important for premature babies who need precise nutrition. Researchers used artificial intelligence and computer optimization to find a better way to mix milk from different donors. Instead of combining milk from just one donor, they tested mixing milk from up to five donors to create batches with more consistent nutrition. The new method worked better than the current approach, getting closer to the target nutrition levels that babies need. This research shows how computers can help hospitals and milk banks provide better nutrition to vulnerable newborns.
The Quick Take
- What they studied: Can computers help milk banks mix donated breast milk from multiple donors to make it more nutritionally consistent for premature babies?
- Who participated: The study analyzed 2,236 individual milk samples from 480 women who donated breast milk to milk banks. Researchers looked at information like the donors’ age, diet, weight, and how long they had been breastfeeding.
- Key finding: When milk from multiple donors (up to 5) was mixed together using computer predictions, the nutrition levels were much more consistent. The new method reduced nutrition variation by about 40% compared to mixing milk from just one donor.
- What it means for you: If you have a premature baby in the hospital receiving donor milk, this research suggests they may receive more consistent nutrition in the future. However, hospitals would need to adopt this new system first, so changes may take time.
The Research Details
Researchers collected data from 480 women who donated breast milk, creating 2,236 individual milk samples. They measured the protein and calorie content of each sample and recorded information about each donor including age, diet type (vegetarian or not), body weight, whether they had a full-term or premature delivery, how long they’d been breastfeeding, and how much milk they pumped.
They then used machine learning—a type of artificial intelligence that learns patterns from data—to predict the nutrition content of milk based on donor information. The computer system tested different combinations of milk from multiple donors to find the best mixes that would match target nutrition levels (1.0 grams of protein and 70 calories per 100 milliliters of milk).
The researchers compared their new multi-donor mixing strategy with the current method used at most milk banks, which combines milk from different batches but only from a single donor. They measured how close each approach came to the target nutrition levels.
Premature babies have very specific nutrition needs, and their bodies can’t handle big changes in what they eat. When donor milk varies too much in nutrition, babies may not get enough protein or calories, which can affect their growth and development. A computer system that can predict milk nutrition and find the best combinations could help hospitals give babies more consistent nutrition, which is especially important for the tiniest and most vulnerable patients.
This study used a large dataset (2,236 samples from 480 donors), which is a strength. The researchers tested their computer predictions and found that random forest regression (a specific type of machine learning) worked best. However, this is a theoretical study—they didn’t actually test the new pooling method in real hospitals yet. The study also didn’t include information about how the new system would work in practice or whether hospitals could actually implement it.
What the Results Show
The machine learning system successfully predicted milk protein and calorie content based on donor characteristics. When the researchers used these predictions to create milk pools from multiple donors (up to 5), the nutrition levels were much more stable than the current single-donor approach.
Specifically, the new multi-donor pooling strategy had an average total absolute deviation of 0.402 from target nutrition levels, compared to 0.664 for the current single-donor method. In simpler terms, the new approach got about 40% closer to the ideal nutrition targets.
The computer system was able to find combinations of milk from different donors that balanced out natural variations. For example, if one donor’s milk was slightly low in protein, the system could mix it with milk from another donor whose milk was slightly high in protein, creating a more balanced final product.
The study identified which donor characteristics were most important for predicting milk nutrition. Factors like maternal age, diet type, and lactation stage (how long the mother had been breastfeeding) all influenced the nutrition content. The research also showed that using data from up to 5 donors provided better results than using fewer donors, but the improvement leveled off after that point.
Most human milk banks currently use a single-donor pooling strategy, combining different batches from the same mother to try to standardize nutrition. This study shows that a multi-donor approach could work better. The use of machine learning for this purpose is relatively new in milk banking, so this research represents an advancement in how technology could be applied to improve donor milk quality.
This study is theoretical—researchers didn’t actually implement the new system in real milk banks to see if it works in practice. They also didn’t test whether mixing milk from multiple donors could create any other issues, such as concerns about traceability or safety protocols. The study focused only on protein and calorie content and didn’t examine other important nutrients like fats or vitamins. Additionally, the research didn’t account for practical challenges like whether milk banks have the technology and training to use such a system, or whether it would be cost-effective.
The Bottom Line
This research suggests that machine learning-based multi-donor pooling could improve the consistency of donor breast milk nutrition. However, this is still a theoretical study. Before hospitals adopt this approach, they would need to test it in real-world settings and ensure it’s safe and practical. If you have a premature baby receiving donor milk, you can ask your hospital whether they’re aware of these new approaches, but don’t expect immediate changes. (Confidence: Moderate—the theory is sound, but real-world testing is needed.)
Parents of premature babies in neonatal intensive care units should care about this research, as it could eventually improve their baby’s nutrition. Milk bank directors and hospital administrators should pay attention because this could improve their operations. Healthcare providers working with premature infants should be aware of this emerging technology. People who donate breast milk may also be interested in how their donations are used.
If hospitals decide to adopt this system, it would likely take several years for widespread implementation. First, the system would need to be tested in real hospitals (1-2 years), then refined based on results (1-2 years), and finally rolled out to other facilities. So realistic changes in hospitals might not happen for 3-5 years or more.
Want to Apply This Research?
- If you’re tracking a premature baby’s nutrition, record weekly measurements of protein and calorie intake from donor milk along with growth metrics (weight gain, length). Compare these measurements month-to-month to see if nutrition consistency improves.
- Parents can ask their hospital’s milk bank about nutrition consistency reports for their baby’s donor milk. Request information about how milk is selected and pooled. If your hospital isn’t using advanced methods, you could ask whether they plan to adopt new technologies to improve milk consistency.
- Track your baby’s growth progress (weight, length, head circumference) at regular check-ups and correlate it with donor milk nutrition data from the hospital. Over time, more consistent nutrition should support more stable growth patterns. Keep records of which donors provided milk and any notes about nutrition content if available.
This research is theoretical and has not yet been tested in actual hospital settings. It should not be used to make medical decisions about your baby’s care. Always follow your pediatrician’s and hospital’s recommendations for feeding premature infants. If you have questions about your baby’s donor milk or nutrition, speak with your neonatal care team. This article is for informational purposes only and is not a substitute for professional medical advice.
This research translation is published by Gram Research, the science division of Gram, an AI-powered nutrition tracking app.