Probing clarity: AI-generated simplified breast imaging reports for enhanced patient comprehension powered by ChatGPT-4o

• DOI: 10.1186/s41747-024-00526-1
• Publication Date: 2024-10-30T17:01:52Z
• Authors: Roberto Maroncelli; Veronica Rizzo; Marcella Pasculli; Federica Cicciarelli; Massimo Macera; Francesca Galati; Carlo Catalano; Federica Pediconi
• Keywords: Breast imaging; Breast MRI

Abstract Background To assess the reliability and comprehensibility of breast radiology reports simplified by artificial intelligence using large language model (LLM) ChatGPT-4o. Methods A radiologist with 20 years’ experience selected 21 anonymized reports, 7 mammography, ultrasound, magnetic resonance imaging (MRI), categorized according to reporting data system (BI-RADS). These underwent simplification prompting ChatGPT-4o “Explain this medical report a patient simple language”. Five radiologists assessed quality these for factual accuracy, completeness, potential harm 5-point Likert scale from 1 (strongly agree) 5 disagree). Another evaluated text comprehension five non-healthcare personnel readers (excellent) (poor). Descriptive statistics, Cronbach’s α, Kruskal–Wallis test were used. Results Mammography, MRI showed high accuracy (median 2) completeness across radiologists, low scores 5); no significant group differences ( p ≥ 0.780), internal consistency (α > 0.80) observed. Non-healthcare 2 mammography ultrasound); modalities = 0.368), 0.85) BI-RADS 0, 1, accurately explained, while 3–6 challenging. Conclusion The demonstrated clarity, offering promise patients diverse backgrounds. LLMs like could simplify aid in communication, enhance care. Relevance statement Simplified generated show enhancing communication patients, improving varying educational backgrounds, contributing patient-centered care practice. Key Points AI simplifies complex understanding. maintain significantly. Implementing enhances engagement imaging. Graphical