ABSTRACT
We conclude section II of the book with a survey of advanced methods as well as emerging and newer innovations in EHR research. The chapter begins within the domain of data science, covering machine learning applications for exploratory analysis, predictive modeling, and knowledge discovery and data mining. Data mining is especially relevant to EHR research to extract and process the abundance of free text found in the unstructured clinical narrative. This free text processing may include string parsing techniques such as regular expressions or rule-based and machine-learning natural language processing. Following the treatment of data science topics, the chapter progresses to advanced epidemiologic study designs and analytic approaches, including quasi-experimental studies, pragmatic clinical trials and target trials, spatial analysis, Bayesian analysis, and qualitative analysis. Finally, the chapter concludes with an overview of phenome-wide association studies from the EHR.
