This document discusses the differences between clinical trials and health outcomes research. Clinical trials use homogeneous samples, surrogate endpoints, and focus on a single outcome. They are also typically underpowered for rare events. Health outcomes research uses heterogeneous data from the general population to examine multiple real endpoints simultaneously. It has larger samples and data that allow analysis of rare occurrences. Predictive modeling is better suited than traditional statistical methods for analyzing heterogeneous health outcomes data due to relaxed assumptions like normality.