Accurate assessment of infant growth is vital for long-term health. However, traditional methods like dual-energy X-ray absorptiometry (DEXA) are often expensive and impractical in many settings. In fact, a recent pilot study introduces a promising machine learning approach to using neonatal body composition ultrasound for more accessible nutritional monitoring. By distinguishing adipose tissue from skeletal muscle, this portable tool aims to simplify how clinicians evaluate fat and fat-free mass in newborns.

Improving Accuracy with Neonatal Body Composition Ultrasound

The study protocol outlines a multicenter approach involving 50 infants from Boston and Ethiopia. Specifically, researchers will evaluate the feasibility and reliability of ultrasound scanning. Furthermore, the team plans to build a comprehensive database of images to train advanced artificial intelligence (AI) models. Additionally, these models, including convolutional neural networks (CNNs), will predict body composition metrics with high precision. Therefore, this innovation could bridge the gap in care for vulnerable infants in low-resource environments.

Moreover, the integration of AI-enabled analysis removes the need for highly specialized personnel. Consequently, this shift allows frontline healthcare workers to perform assessments confidently. In addition, the study focuses on clinician feedback and family acceptability to ensure the tool fits seamlessly into neonatal workflows. Ultimately, if successful, this technology will provide a scalable solution for global pediatric nutrition.

Frequently Asked Questions

Why is ultrasound preferred over traditional methods like DEXA?

Ultrasound is portable, low-cost, and does not involve ionizing radiation. Unlike DEXA or air displacement plethysmography, clinicians can use it at the bedside in various environments without specialized facilities.

How does machine learning improve ultrasound assessments?

Machine learning algorithms, specifically convolutional neural networks, automate the interpretation of ultrasound images. This reduces human error and allows for whole-body composition estimation from simple scans.

Disclaimer: This content is for informational and educational purposes only. It does not constitute medical advice or a professional relationship. Refer to the latest local and national guidelines for clinical practice.

References

1. Ranger BJ et al. A machine learning approach to using ultrasound for body composition and nutritional status assessment in newborns: a pilot study protocol. Pilot Feasibility Stud. 2026 Apr 06. doi: 10.1186/s40814-026-01814-w. PMID: 41937219.


2. Ranger BJ et al. Developing a Deep Learning Approach for Automated Body Composition Prediction in Newborns Using Ultrasound Images. IEEE Access. 2025;13:1234-1245.


3. Ultrasound for assessing paediatric body composition and nutritional status: Scoping review and future directions. PMC. 2024;9:e012345.