# Breastfeeding Patterns, Chemical Pathology, and Antibiotic Resistance in Lactating Mothers: A Spatial Analysis of Nutritional, Toxicological, and Antimicrobial Implications

## TL;DR

This study explores how breastfeeding patterns, milk composition, and antibiotic resistance are linked in lactating mothers, using spatial analysis and machine learning to identify regional health risks.

## Contribution

The novel contribution is the integration of spatial analysis and machine learning to assess breastfeeding, metabolic markers, and antibiotic resistance in lactating mothers.

## Key findings

- Exclusive breastfeeding is associated with better milk quality and lower antibiotic resistance (18.2%) compared to non-exclusive groups (28.6%).
- Geographically weighted regression highlights regional disparities in breastfeeding exclusivity effects.
- Cholesterol and BMI are top predictors of antibiotic resistance and nutritional status in machine learning models.

## Abstract

This study investigates the interplay between breastfeeding patterns, chemical pathology, and antibiotic resistance in lactating mothers. A cross-sectional analysis was conducted on 1,200 lactating mothers aged 18 to 45, examining breastfeeding practices, biochemical markers, milk composition, and antibiotic resistance status. The findings reveal significant metabolic variations, with mean glucose and cholesterol levels at 135.23 mg/dL and 224.58 mg/dL, respectively, suggesting potential cardiovascular risks. Exclusive breastfeeding improved milk quality by having higher average fat content (3.48%) and lactose (6.96%), and the reported antibiotic resistance was lower (18.2%), compared with non-exclusive groups (28.6%). Geographically weighted regression (GWR) revealed spatial variability in exclusivity effects, highlighting regional nutritional disparities. Machine learning models, random forest, support vector machine (SVM), and gradient boosting machine (GBM), were used to predict resistance and nutritional status, with cholesterol and BMI emerging as the top predictors. Although model performance was modest (AUC ≈ 0.65), random forest achieved moderate discriminative power (AUC ~0.65), with cholesterol and BMI ranked highest in feature importance. Receiver operating characteristic (ROC) analysis for GBM and SVM also indicated moderate predictive capacity. Spatial mapping of antibiotic resistance revealed clustered patterns, emphasizing the need for region-specific interventions. Furthermore, systolic blood pressure showed a weak correlation with cholesterol levels, indicating independent metabolic risks. This study underscores the critical need for integrated nutritional and antimicrobial stewardship in lactating mothers, particularly in regions with identified spatial vulnerabilities. Policy implications suggest targeted nutritional support and regional antibiotic surveillance to mitigate health risks in this population.