Retrospective patient cohort analysis is a widely used technique in many healthcare studies. Due to its data intensive nature, the traditional analytical pipeline requires expertise from several areas, such as databases, data mining, software development, statistics, and domain knowledge. As a result, domain experts often rely on a team of technologists to help perform such studies which can make the process slow and cumbersome. To allow domain experts to perform faster and more flexible analyses, we designed an integrated system that combines visual exploration and data analytics with an intuitive user interface. Our system lets clinicians interactively visualize and refine cohorts, request analytics on those cohorts, and make new discoveries.