Exploring the role of breastfeeding, antibiotics, and indoor environments in preschool children atopic dermatitis through machine learning and hygiene hypothesis.

The increasing global incidence of atopic dermatitis (AD) in children, especially in Western industrialized nations, has attracted considerable attention. The hygiene hypothesis, which posits that early pathogen exposure is crucial for immune system development, is central to understanding this trend. Furthermore, advanced machine learning algorithms have provided fresh insights into the interactions among various risk factors. This study investigates the relationship between early childhood antibiotic use, the duration of exclusive breastfeeding, indoor environmental factors, and child AD. By integrating machine learning techniques with the hygiene hypothesis, we aim to assess and interpret the significance of these risk factors. In this community-based case-control study with a 1:4 matching design, we evaluated the prevalence of AD in preschool-aged children. Data were collected via questionnaires completed by the parents of 771 children diagnosed with AD, matched with controls based on gender, age, and ethnicity. Univariate analyses identified relevant characteristics, which were further examined using multivariable logistic regression to calculate odds ratios (ORs). Stratified analyses assessed confounders and interactions, while the significance of variables was determined using a machine learning model. Renovating the dwelling during the mother's pregnancy (OR = 1.50; 95% CI 1.15-1.96) was identified as a risk factor for childhood AD. Additionally, antibiotic use three or more times during the child's first year (OR = 1.92; 95% CI 1.29-2.85) increased the risk of AD, independent of the parents' history of atopic disease and the child's mode of birth. Moreover, exclusive breastfeeding for four months or more (OR = 1.59; 95% CI 1.17-2.17) was identified as a risk factor for AD, particularly in the group without a maternal history of atopic disease. In contrast, having older siblings in the family (OR = 0.76; 95% CI 0.63-0.92) and low birth weight (OR = 0.62; 95% CI 0.47-0.81) were identified as protective factors against AD. Machine learning modeling indicated that the duration of exclusive breastfeeding, having older siblings, low birth weight, and parental history of AD or allergic rhinitis are key predictors of childhood AD. Our findings support the broader interpretation of the hygiene hypothesis. Machine learning analysis highlights the key role of the hygiene hypothesis and underscores the need for future AD prevention and healthcare initiatives focusing on children with a parental history of AD or allergic rhinitis. Moreover, minimizing antibiotic overuse may be essential for preventing AD in children. Further research is necessary to elucidate the impact and mechanisms of exclusive breastfeeding on AD to instruct maternal and child healthcare practices.