Detecting Adverse Drug Events from Swedish Electronic Health Records Using Text Mining

摘要

Electronic Health Records are a valuable source of patient information which can be leveraged to detect Adverse Drug Events (ADEs) and aid post-mark drug-surveillance. The overall aim of this study is to scrutinize text written by clinicians in Swedish Electronic Health Records (EHR) and build a model for ADE detection that produces medically relevant predictions. Natural Language Processing techniques are exploited to create important predictors and incorporate them into the learning process. The study focuses on the five most frequent ADE cases found in the electronic patient record corpus. The results indicate that considering textual features, can improve the classification performance by 15% in some ADE cases, compared to a method that used structured features. Additionally, variable patient history lengths are included in the models, demonstrating the importance of the above decision rather than using an arbitrary number for a history length. The experimental findings suggest the importance of the variable window sizes as well as the importance of incorporating clinical text in the learning process, as it is highly informative towards ADE prediction and can provide clinically relevant results.