Integrating GPT-4o for Data Mining in Neurosurgery: A Feasibility and Proof-of-Concept Study (Preprint)

Abstract

<sec> <title>UNSTRUCTURED</title> The integration of large language models (LLMs) like GPT-4o in healthcare offers transformative potential in data mining, especially for unstructured medical records. This proof-of-concept study evaluated GPT-4o’s accuracy in extracting structured data from raw neurosurgical reports of patients with vestibular schwannoma, achieving up to 100% accuracy for several data categories after prompt refinement. The findings underscore the value of AI in facilitating clinical research and operational efficiencies, while highlighting privacy and data security concerns. This discussion explores the broader implications of LLMs for healthcare data mining, operational predictions, and automated medical coding, emphasizing the need for robust data governance frameworks. </sec>

Ähnliche Arbeiten