AI Algorithm Improves Intravenous Nutrition for Premature Babies, Stanford Study Finds

AI Revolutionizes Neonatal Nutrition

A groundbreaking study from Stanford Medicine has demonstrated that artificial intelligence (AI) can significantly improve the process of prescribing intravenous (IV) nutrition for premature babies. Published in Nature Medicine on March 25, 2025, the research highlights how AI algorithms can enhance clinical decision-making for vulnerable newborns 1

1

2

2

3

3

.

Current Challenges in Neonatal Nutrition

Premature babies, particularly those born more than eight weeks early, often require IV nutrition as their digestive systems are not mature enough to absorb nutrients. This process, known as total parenteral nutrition (TPN), is currently the largest source of medical errors in neonatal intensive care units globally 1

1

4

4

.

Dr. Nima Aghaeepour, senior study author and associate professor at Stanford, explains:

"Right now, we come up with a TPN prescription for each baby, individually, every day. We make it from scratch and provide it to them." 1

1

The current method is not only error-prone but also time-consuming, requiring input from six experts in a multi-hour process 2

2

4

4

.

AI-Powered Solution

The Stanford team developed an AI algorithm trained on a decade of electronic medical records from Lucile Packard Children's Hospital Stanford. This included 79,790 TPN prescriptions from 5,913 premature patients 1

1

2

2

3

3

.

Key features of the AI solution include:

1. Prediction of nutrient needs based on patient data
2. Grouping of similar prescriptions into 15 standard formulas
3. Daily adjustment of recommendations as patients grow and conditions change 1

   1

   2

   2

   3

   3

Promising Results

The AI-generated prescriptions showed impressive performance when tested against human-created ones:

1. In a blind test, 10 neonatologists consistently preferred the AI-generated prescriptions 5

   5

   .
2. Analysis of past patient records revealed that significant deviations from AI recommendations were associated with higher risks of death, sepsis, and bowel disease 5

   5

   .

Dr. Shabnam Gaskari, co-author and chief pharmacy officer at Stanford Medicine Children's Health, notes:

"If we had manufactured, ready-to-use TPNs, that would be very beneficial. I think it would be safer for patients." 2

2

Future Implications

The researchers plan to conduct a clinical trial comparing outcomes between babies fed using traditional methods and those using AI-recommended nutrition 5

5

. If successful, this approach could:

1. Reduce medical errors
2. Save time and money
3. Make preemie care more accessible in low-resource settings
4. Allow healthcare providers more time for patient interaction 1

   1

   5

   5

Dr. David Stevenson, a neonatologist and study co-author, concludes:

"This reflects our hope for how AI will enhance medicine: What it's going to do is make doctors better and make top-notch care more accessible." 5

5

As AI continues to evolve in healthcare, this study represents a significant step forward in improving care for some of the most vulnerable patients.