Improving the sensitivity of the problem list in an intensive care unit by using natural language processing

Area of Science:

• Medical Informatics
• Clinical Documentation Improvement
• Natural Language Processing in Healthcare

Background:

• Electronic problem lists are crucial for patient care but often incomplete.
• Manual review of clinical notes for problem list updates is time-consuming and error-prone.

Purpose of the Study:

• To develop and evaluate a Natural Language Processing (NLP) system for automatically extracting potential medical problems from clinical free-text documents.
• To assess the impact of this NLP system on the completeness of electronic problem lists in an intensive care unit (ICU).

Main Methods:

• A prospective randomized controlled trial was conducted with 105 ICU patients.
• Patients were assigned to either a control group or an intervention group.
• In the intervention group, an NLP system analyzed clinical documents and proposed potential medical problems for inclusion in the electronic problem list.

Main Results:

• The NLP system significantly increased the sensitivity of electronic problem lists from 8.9% to 41%.
• When automatically proposed problems, even if not acknowledged by users, were included, sensitivity rose to 77.4%.
• This demonstrates a substantial improvement in capturing patient medical issues.

Conclusions:

• Automated extraction of medical problems using NLP is effective in improving electronic problem list completeness.
• The developed NLP system shows significant potential for enhancing clinical documentation and patient care in critical care settings.
• Further integration and refinement of such systems can optimize problem list management.