Leveraging Generative Pre-Trained Transformer Models for Standardizing Nursing Data

Care plans documented by nurses in electronic health records (EHR) are a rich source of data to generate knowledge and measure the impact of nursing care. However, these structured data are often documented using non- standardized local terms, which hinders interoperability and wide- scale analysis of these data. In this paper, we compared the performance of OpenAl's Generative Pre-Trained Transformer models (GPT -3.5 and GPT -4), accessed through the ChatGPT web application and Application Programming Interface (API) platforms, in mapping local care plan terms to standardized nursing terms. Our results demonstrate that the GPT -4 API model outperformed others by minimizing hallucinations, offering enhanced accuracy, and yielding a higher percentage of matched terms. GPT models are valuable tools in advancing the standardization and interoperability of nursing data.