Search

In this paper, the authors describe a method for exploring the feasibility of using Natural Language Processing (NLP) and Machine Learning (ML) techniques to analyze patient safety incident database reports for themes. We developed a novel thematic analysis strategy to automatically detect keywords and latent themes that describe HIT-related patient safety incidents. The strategy was applied to patient safety reports to test the approach. The efforts by the automated strategy were compared to the efforts by analysts who manually reviewed and identified key words, topics, and themes for the same reports. The computer-based error themes were also compared to the human-determined themes for crosschecking. The manual thematic analysis took about 150 hours to complete on the patient safety reports. The semi-automated approach took only 10% of that time. 95% of the themes extracted from the automated method were aligned with the themes from the manual process. The findings underscore the utility of NLP and ML in identifying thematic patterns embedded in large numbers of unstructured data. The NLP-ML method therefore represents a valuable addition to the tools of detecting and understanding HIT-related errors.

Correction Unfortunately, the original article [1] contained an error. The additional files were included incorrectly. The correct additional files 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13 and 14 are published in this correction.