The increasing availability of Big Data in healthcare encourages investigators to seek answers to big questions. However, nonparametric approaches to analyzing these data can suffer from the curse of dimensionality, and traditional parametric modeling does not necessarily scale. Targeted learning (TL) combines semiparametric methodology with advanced machine learning techniques to provide a sound foundation for extracting information from data. Predictive models, variable importance measures, and treatment benefits and risks can all be addressed within this framework. TL has been applied in a broad range of healthcare settings, including genomics, precision medicine, health policy, and drug safety. This article provides an introduction to the two main components of TL, targeted minimum loss-based estimation and super learning, and gives examples of applications in predictive modeling, variable importance ranking, and comparative effectiveness research.