Speakers - WNSC 2025

Abstract

The integration of artificial intelligence (AI) in medical research has notably advanced the analysis of complex biological systems, including human milk (HM), which is essential for infant development. Despite significant progress, the detailed nutritional profiling of HM utilizing machine learning (ML) has been limited and remains a promising area of research.

Objective: To systematically explore and synthesize the literature on the application of ML in analysing HM to determine its macro- and micronutrient compositions.

Methods: This scoping review was conducted following the method and protocol outlined by the Joanna Briggs Institute Methods Manual. Four databases—PubMed, CINAHL, SCOPUS, and ACM Digital Library—were searched in March 2024. Screening and data extraction were conducted using predefined inclusion criteria and the PRISMA-Scr framework to ensure methodological rigor.

Results: This review encompassed five articles published between 2021 and 2024, employing cross-sectional (n = 4) and cohort study designs (n = 1). Sample sizes varied, ranging from six to over a thousand breast milk samples. Collection methods included pumping, direct expression, and donation from milk banks. ML techniques, such as linear and non-linear ML algorithms, were diversely applied across studies to predict various milk components, including macronutrients, micronutrients, minerals, and hormones. These studies collectively highlight the diverse applications of predictive modeling and innovative methodologies in understanding human milk composition and its implications for infant health and development.

Conclusion: This scoping review underscores the significant potential of machine learning in advancing our understanding of HM composition and its impact on infant health. The findings suggest that ML techniques can effectively assist in data fitting across various analytical methodologies, offering a promising approach for comprehensive analysis. Machine learning shows substantial promise in enhancing both comprehension and predictive accuracy within diverse methodological frameworks. Therefore, ongoing investigation into the application of machine learning in human milk research is essential for thorough analysis and for unravelling the complexities of human milk and its implications for infant health.