Investigating Clinical Named Entity Recognition Approaches for Information Extraction from EMR

Abstract

AbstractElectronic Medical Record (EMR) contains much information used in various applications, such as identifying similar patients, keeping track of follow-ups, etc. An essential feature of EMR is that it is rich in context and may lead to ambiguity during analysis if undetected in the initial stages and could result in wrong interpretation. The chapter includes a detailed literature review of recent clinical Named Entity techniques. The chapter demonstrates comparative results of Clinical Named Entity Classification using rule-based, deep learning-based, and hybrid approaches. The chapter expresses the efficacy of clinical Named Entity Recognition (NER) techniques for Information Extraction. Our experimentation validates state-of-art recitation about the high accuracy of combined Deep Learning (DL) models with a sequential model. The experiment appraises the need for improved clinical word embeddings for efficient entity identification.KeywordsInformation Extraction (IR)Electronic Medical Record (EMR)(Clinical Named Entity Recognition) cNERDeep Learning (DL)Long and Short Term Memory (LSTM)Bidirectional LSTM (BLSTM)