Natural Language Processing for Health-Related Texts

Narrative text is an important component of communication in health care, including patient-specific information in health record reports and notes and general biomedical knowledge papers, textbooks and web resources. Retrieval of information from such sources can be accomplished with keyword indexing but this approach fails to distinguish between texts that discuss a topic versus those that merely mention it. Natural language processing (NLP) techniques seek to analyze narrative text to make these distinctions and even to find instances where a concept is discussed but not explicitly mentioned. Previously, grammar-based methods were based on parsing sentences into their structure to infer semantics. Machine-learning methods are now being applied that can infer semantics through techniques such as statistical pattern recognition. Regular expression, rule bases, neural networks and word embeddings are among approaches that are improving the ability of automated systems to carry out successful language understanding, information extraction and question answering.