Overview of the Potentially Transforming Role of Large Language Models (LlMs) in Creating Patient-Friendly Summaries for Radiology Reports

Author(s): Sadhana Kalidindi, RV Prasanna Vadana

Radiology reports typically contain technical terminology and medical jargon that is generally confusing to patients. Large language models such as Generative Pre-Trained Transformer 4 (GPT-4) (Open AI, San Francisco, USA) could play an effective role in bridging the gap between patient comprehension and radiology reports. With the goal of demystifying all of the medical jargon in radiology reports and encouraging patient involvement and understanding, this review paper investigates the possible function of GPT-4 in producing user-friendly summaries. This article demonstrates the capacity of GPT-4 to accurately interpret and simplify various radiological findings in a way that even someone who is not familiar with technical terms can understand. It evaluates how these developments affect treatment plan adherence, patient satisfaction, and overall health outcomes. Additionally, the paper explores the possible drawbacks and moral dilemmas related to applying AI-driven summaries in clinical practice, such as accuracy, privacy, and the requirement for human supervision.