Leveraging Large Language Models for Medication Named Entity Recognition and Text Expansion in Multilingual Clinical Data

Abstract

Medication prescriptions found in hospital discharge summaries are often presented in free-text form, which lacks consistency and structure. These texts frequently contain a mixture of local languages, non-standard abbreviations, and brand-specific drug names, making them difficult to interpret accurately. The unstructured nature of these statements poses significant challenges for both healthcare professionals and automated systems, particularly when it comes to ensuring patient safety and facilitating effective clinical decision-making. With the rapid advancement of artificial intelligence, large language models (LLMs) such as ChatGPT3.5 have demonstrated impressive capabilities in understanding and generating human-like text across a variety of domains. These models offer a promising approach to addressing the complexities of medical text by enabling automatic extraction and clarification of critical information. In this study, we leverage the capabilities of ChatGPT-3.5 to process and refine medication statements from discharge summaries. Our goal is to automatically identify key medical entities and expand these statements into more structured and interpretable formats. By doing so, we aim to enhance the readability and usability of medication information for both humans and machines, ultimately contributing to improved healthcare data management and patient care.

Ähnliche Arbeiten