Patient safety is a fundamental aspect of the quality of healthcare and there is a growing interest in improving safety among healthcare stakeholders in many countries. The Korean government recognized that patient safety is a threat to society following several serious adverse events, and so the Ministry of Health and Welfare of the Korean government set up the Patient Safety Act in January 2015. This study analyzed text data on patient safety collected from web-based, user-generated documents related to the legislation to see if they accurately represent the specific concerns of various healthcare stakeholders. We adopted the unsupervised natural language processing method of probabilistic topic modeling and also Latent Dirichlet Allocation. The results showed that text data are useful for inferring the latent concerns of healthcare consumers, providers, government bodies, and researchers as well as changes therein over time.
Keywords: Patient safety; healthcare stakeholders; natural language processing; topic modeling.