UCLA

This dissertation develops and applies advanced statistical and machine learning methods to address challenges in biomedical data analysis. The work spans six interconnected studies that contribute methodological innovations and practical applications across epidemiological surveillance, longitudinal modeling, clinical prediction, and precision medicine.Chapter 1 presents efficient MCMC algorithms for Bayesian changepoint detection in overdispersed autoregressive count data, with applications to COVID-19 surveillance. Chapter 2 introduces penalized fractional polynomials via mixed models for flexible non-linear longitudinal data analysis. Chapter 3 compares machine learning models for hospital readmission prediction. Chapter 4 optimizes blood glucose predictions using advanced forecasting techniques. Chapter 5 develops a generative latent protocol model for sepsis treatment in ICU settings. Chapter 6 presents an explainable machine learning model for hypertension screening in Type 1 diabetes patients using continuous glucose monitoring data.Together, these studies demonstrate the power of combining classical statistical theory with modern computational methods to solve pressing problems in biomedical research and clinical practice.