Discovering and Applying Knowledge in Clinical Databases

点击翻译按钮获取中文摘要

The long term goal of our ongoing project, "Discovering and applying knowledge in clinical databases," is to learn from data in the electronic health record (EHR) and to apply that knowledge to relevant problems. The increasing adoption of the EHR promises to provide data for clinical research and informatics research, but secondary use of the data has been limited. Challenges include the complexity, incompleteness, and inaccuracy of the record. We propose to study the EHR from an information theoretic point of view, treating the EHR as a natural object worthy of study, and applying methods from non-linear time series analysis. Armed with a better understanding of the record, we hope to measure and account for data completeness and to improve interpretation and use of the data. We hypothesize that we can characterize an electronic health record using a formal information theoretic framework, and that the measured properties can help answer informatics and clinical questions. Our aims are to (1) develop an information theoretic framework for characterizing the electronic health record, (2) use the information theoretic framework to study EHR and sampling issues, and (3) use the framework and traditional data mining to answer clinical and informatics questions. We will approach the EHR as a complex time series and characterize the information in the record using univariate sequential mutual information (the degree to which observations of a variable predict future observations) and a network of pair-wise mutual information among all variables, discrete and continuous. The result will be a measure of the predictability of the record and a set of associations among clinical features. We will use the predictability results to study the completeness of a patient's record, the appropriateness of a clinician's sampling rate, outlier data points, and changes in patient acuity. We will use predictability and associations to link narrative abstractions with their primary data, to interpret narrative modifiers, to cluster terms, to find associations (in the context of phenome-wide association studies), and to carry out exploratory analyses of defining phenotype profiles and of mutual information-based surveillance.