This chapter extends the statistical modeling approaches introduced in earlier chapters by focusing on supervised machine learning techniques for probabilistic classification and prediction. We introduce three foundational models such as Naive Bayes classifiers, Bayesian Networks, and Bayesian Additive Regression Trees (BART), each offering distinct strategies for handling structured data in clinical and epidemiological research. Naive Bayes models rely on conditional independence assumptions to enable fast and interpretable predictions, while Bayesian Networks represent joint distributions through directed acyclic graphs, allowing the incorporation of expert knowledge and complex dependencies. BART offers a nonparametric alternative capable of capturing nonlinear relationships and interactions through an ensemble of decision trees. For each method, we discuss theoretical underpinnings, estimation procedures, and model diagnostics, alongside applications to infection prediction using structured health data. Emphasis is placed on uncertainty quantification, model calibration, and comparative evaluation. These models offer flexible extensions to the generalized linear modeling framework and serve as a foundation for more advanced Bayesian and causal modeling techniques.