The Use Of Natural Language Processing In Extracting Information From Medical Records

Abstract

Utilizing natural language processing (NLP) to analyze symptoms extracted from electronic health records (EHRs) has the potential to enhance the progress of symptom research. Our objective is to consolidate the existing research on the use of Natural Language Processing (NLP) in the processing and analysis of symptom information found in free-text narratives inside Electronic Health Records (EHR). For each research, information pertaining to the objective, free-text collection, patients, symptoms, NLP approach, assessment metrics, and quality indicators were retrieved. Fourteen studies reported symptom-related information as their main endpoint. The EHR narratives included a range of therapeutic disciplines, including general, cardiology, and mental health, with the latter being the most common. The studies examined a broad range of symptoms, such as dyspnea, pain, emesis, vertigo, disrupted sleep, constipation, and low mood. Natural Language Processing (NLP) methodologies included pre-existing NLP tools, categorization techniques, and meticulously maintained rule-based processing. Natural Language Processing (NLP) is used to retrieve data from Electronic Health Record (EHR) unstructured narratives authored by a multitude of healthcare practitioners, including a wide spectrum of symptoms across many clinical domains. The present emphasis in this research is on developing techniques to collect symptom data and use that data for illness categorization purposes, rather than studying the symptoms themselves. Future research in Natural Language Processing (NLP) should focus on examining symptoms and the documenting of symptoms in electronic health record (EHR) narratives that are written in free-text format. It is important to make an effort to analyze the features of patients and create publically accessible NLP algorithms or pipelines and vocabularies that are specifically designed for symptoms.