Natural Language Processing of Sentiments Identified in Patient Comments Associated with Less Than Top-Rated Care

Area of Science:

• Healthcare Informatics
• Natural Language Processing
• Patient Experience Research

Background:

• Patient feedback is crucial for healthcare improvement initiatives.
• Natural language processing (NLP) offers a method to analyze qualitative patient comments.
• Understanding patient sentiment and topics can guide quality enhancement.

Purpose of the Study:

• To quantify sentiments and identify themes in patient comments linked to suboptimal experience ratings.
• To explore the relationship between specific linguistic features and patient satisfaction levels.
• To leverage NLP for actionable insights into patient care quality.

Main Methods:

• Analysis of 1117 patient comments associated with ratings from 1 to 4 (out of 5).
• Sentiment analysis using Linguistic Inquiry and Word Count software.
• Topic modeling to identify prevalent themes within the comments.

Main Results:

• Positive sentiments correlated with higher numerical ratings.
• Negative tones, word count, numbers, and ethnicity mentions were associated with lower ratings.
• Specific topics emerged: 'listening, concern, collaboration' for 1-star ratings and 'logistics, pain' for 4-star ratings.

Conclusions:

• NLP analysis of patient comments aligns with existing evidence on care experience factors.
• Worst ratings are linked to interpersonal/relationship issues; moderate ratings to process issues.
• NLP effectively analyzes large volumes of patient feedback to pinpoint areas for care improvement.