Unstructured clinical narratives can be found in large volumes within the Electronic Health Records (EHRs) such as physician notes, discharge summaries, and radiology reports. Structured data in EHRs are useful in providing a standard analysis, unstructured text can provide rich contextual information that is essential in future, advanced clinical decision-making. The Natural Language Processing (NLP) has become one of the game-changing tools to derive meaningful information out of such disorganized healthcare data. The present paper reviews NLP-based clinical information extraction pipeline including the preprocessing, named entity recognition (NER), extraction of relationships, and concept normalization. We test the system on a real-life EHR data, showing that it is very effective in finding conditions, medications and procedures. Our results suggest that the NLP strategies can have the potential to support the clinical workflow through data-driven decision support and drive the promise of precision medicine.
Veerendra Nath Jasthi (Fri,) studied this question.