A Machine Learning-Based Model for Predicting Breast Milk Flora Ethnicity

Maternal milk is a significant source of nutrients for the infant and is rich in microbial resources, and the maternal microbiota is now considered an essential determinant of infant health that specific perinatal factors may influence. As a social group distinguished from other populations based on the perception of common ancestry, common culture, and common history, different ethnic groups have their local characteristics, living habits, diets, and cultures, so there may be some differences in the human breast milk microbiota of each ethnic group. Based on this, this paper proposes using machine learning and data mining to construct a model to predict ethnicity. We first obtained more convincing sample features by annotating and merging Operational Taxonomic Units (OTU), constructed four machine learning-based ethnicity prediction models using these features and the best model parameters obtained by the grid search method, We established a baseline model to compare the models. The results show that after fivefold cross-validation and comparative analysis, the random forest can effectively predict ethnicity based on breast milk flora with accuracy, precision, recall, and F1 of 0.852, 0.82, 0.836 and 0.823, respectively, which is better than other machine learning models. Therefore, the method proposed in this paper can accurately predict ethnicity, proving the variability of breast milk microbiota among ethnic groups, and providing a solid theoretical basis for future precision diet and precision medicine.