The role of machine learning in real-world evidence generation

This chapter discusses the role of machine learning (ML) and other computationally intensive tools and their use in RWD. Machine learning can be thought of as a collection of computational methods or algorithms that can be used to generate predictions or to cluster observations into similar subgroups. These tools can be incorporated into analysis of RWD to ascertain patterns, identify homogeneous subgroups, and, with appropriate considerations, to improve efficiency and reduce bias in comparative effectiveness and safety research. This chapter considers problems and methods in which machine learning can potentially improve inference, and others in which the use of ML should be avoided.