A Comparison of Bayesian and Frequentist Methods for Identifying Markers of Susceptibility to Asthma

The assessment of patterns of antibiotic use in early life may have major implications for understanding the development of asthma. This paper compares a classical generalized latent variable modelling framework and a Bayesian machine learning approach to define latent classes of susceptibility to asthma based on patterns of antibiotic use in early life. We compare the potential advantages of each method for elucidating clinically meaningful phenotypes or classes.