Using Exploration and Learning for Medical Records Search: An Experiment in Identifying Cohorts for Comparative Effectiveness Research

Abstract : This paper describes an experiment performed on a medical record data set, using an information retrieval (IR) tool that applies the techniques of exploration and learning, to assist a researcher in identifying the most relevant cohorts. The paper presents some brief background on exploration and learning, how they are incorporated in the IR tool, and an instantiation of exploration and learning used for selecting cohorts for a research population. The research problem addressed in this paper is the TREC 2012 Medical Track task: How to provide content-based access to free-text fields of electronic medical records? The stated goal of the task is to find a population over which comparative effectiveness studies can be done.