NLP Falls Short in Healthcare Advancement
By Jan Sevcik  |  Apr 05, 2022
NLP Falls Short in Healthcare Advancement
Image courtesy of and under license from
Debate swirls about how natural language processing can help advance artificial intelligence in the healthcare sector, where most data are unstructured. Jan Sevcik takes a closer look at NLP, the analysis of unstructured text, and how to understand its strategy.

CHATTANOOGA, TENNESSEE - There is a lot of discussion about how natural language processing (NLP) can help advance artificial intelligence (AI), especially in healthcare where most data are unstructured. What is NLP? NLP is a technology that can understand free text, but it is only the first step in utilizing the value of unstructured data. This process is sometimes referred to as converting unstructured data to structured data, but this in itself is not enough to make the output useful for research or other purposes. Free text electronic health record data consists of patient progress notes, emergency department notes, and imaging and radiology results.

There are many approaches to NLP. Traditional approaches are based on rules, grammar, and ontology, but these methods have significant limitations, especially with medical text. This is because medical text consists of incomplete sentence structures, plus these methods do not benefit from continuous learning. Such traditional methods were developed originally for use on newspaper or magazine quality text, and these ontology-based approaches face limitations because the ontologies must be curated and risk becoming out of date. Also, the curation process itself is limited to the knowledge of the curators, rather than utilizing a data-focused approach.

In the last several years, new and more robust NLP strategies have emerged, which are mathematically based and offer many benefits over traditional methods. These mathematical models utilize machine learning (ML) techniques that ingest training data instead of handwritten rules or ontologies, which means that these methods can be trained on less than perfect data. Among other differences, these modern strategies do not require the assigning of parts of speech or many of the other hallmarks of traditional NLP. This is because words that are modifiers can be determined from a mathematical perspective. As for words that can be either a verb o

The content herein is subject to copyright by The Yuan. All rights reserved. The content of the services is owned or licensed to The Yuan. The copying or storing of any content for anything other than personal use is expressly prohibited without prior written permission from The Yuan, or the copyright holder identified in the copyright notice contained in the content.
Continue reading
Buy this article for only US$1.99
- or -
Continue with Linkedin Continue with Google
Share your thoughts.
The Yuan wants to hear your voice. We welcome your on-topic commentary, critique, and expertise. All comments are moderated for civility.