When we think of machine learning, we usually think of the data as having a structured format. Structured data is data that can be organized into rows and columns having discrete values. Much of the patient data in an electronic health record may be stored in, or converted to, this format. In healthcare, individual patients or encounters often form the rows (or observations), and various features (for example, demographic variables, clinical characteristics, lab observations) of the patient/encounter form the columns. Such a format is particularly conducive to performing machine learning analyses using various algorithms.