Natural language processing in the intensive care unit: A scoping review

Area of Science:

• Artificial Intelligence
• Clinical Informatics
• Natural Language Processing

Background:

• Intensive care units (ICUs) generate vast amounts of text data.
• Natural Language Processing (NLP) applications are underutilized in critical care research and quality improvement.
• NLP offers potential for analyzing clinical text data.

Purpose of the Study:

• To review the current applications of NLP in the intensive care specialty.
• To promote understanding of NLP's future clinical potential in critical care.
• To synthesize findings from recent NLP research in intensive care.

Main Methods:

• A scoping review methodology was employed.
• A systematic search of the PubMed database was conducted for articles from the last 10 years.
• Data extraction and narrative synthesis were performed by independent reviewers.

Main Results:

• Eighty-seven articles were included in the review.
• The most common NLP applications involved predicting clinical outcomes (e.g., mortality) and identifying specific clinical concepts (e.g., sepsis).
• Most studies focused on algorithm development and internal validation, with limited clinical implementation.

Conclusions:

• NLP has diverse applications within the ICU, including outcome prediction and concept identification.
• Increased clinician awareness of NLP techniques can foster the development of clinically relevant algorithms.
• Further implementation of NLP tools in clinical settings is warranted to realize their full potential.