In the era of Electronic Health Records (EHR), it is possible to examine the outcomes of decisions made by doctors during clinical practice to identify patterns of care-generating evidence from the collective experience of millions of patients. We will discuss methods that transform EHR data into a de-identified, temporally ordered, patient-feature matrix. We will review use cases, which use the resulting de-identified data, to discover hidden trends, build predictive models, and drive comparative effectiveness studies in a learning health system. We will also discuss the notion of an "Informatics consult" service to make use of such practice-based evidence in clinical care.