10.3389/fpsyg.2012.00132

Vollständiger Abstract

Worum geht es in dieser Arbeit?

<h4>Background</h4>Vaccination is an important public health intervention across the life course but its determinants may differ by life stage. We aimed to identify and compare predictors of childhood vaccination and recent influenza vaccination among older adults using conventional and machine learning approaches.<h4>Methods</h4>We used data from the second wave (2019-2021) of the Brazilian Longitudinal Study of Aging (ELSI-Brazil), a nationally representative cohort of individuals aged 50+. Analyses included 9,211 participants for childhood vaccination and 9,863 for influenza vaccination. Predictors were selected from sociodemographic, environmental, lifestyle, and health domains. Multivariable logistic regression and machine learning algorithms were used, with performance evaluated via area under the receiver operating characteristic curve (AUC).<h4>Results</h4>The best-performing machine learning model demonstrated acceptable performance for childhood vaccination (AUC = 0.72, 95% CI: 0.69-0.74) and recent influenza vaccination (AUC = 0.67, 95% CI: 0.64-0.70). For childhood vaccination, both analytical approaches identified age, sex, rural residence, parental education, school attendance at age 10, childhood access to books, and allergy as important predictors. Machine learning additionally identified age at school initiation, childhood economic status, self-rated childhood health, severe diarrhea, and history of childhood infectious diseases. Key predictors for recent influenza vaccination included age, marriage, educational level, employment, and hypertension/diabetes. Machine learning additionally highlighted the importance of vaccination history during childhood, life satisfaction, BMI, and mental health.<h4>Conclusion</h4>Early-life socioeconomic and health conditions are important predictors of childhood immunization history, whereas current sociodemographic and health status are key predictors of recent vaccination. Machine learning identified supplementary predictors beyond traditional methods.

Abstract: PubMed · Datensatz