Improving Patient Communication by Simplifying AI-Generated Dental Radiology Reports With ChatGPT: Comparative Study.

Area of Science:

• Medical communication
• Health informatics
• Artificial intelligence in healthcare

Background:

• Radiology reports are often complex and difficult for patients to understand.
• This communication gap can hinder patient engagement and lead to misinterpretations.
• Artificial intelligence (AI), particularly large language models, offers potential solutions for simplifying medical documentation.

Purpose of the Study:

• To evaluate if AI-generated radiology reports simplified by ChatGPT enhance patient understanding, readability, and communication quality.
• Comparison between original AI-generated reports and simplified versions.

Main Methods:

• Three versions of AI-generated radiology reports were created using ChatGPT: original, simplified, and accessibility-optimized.
• 300 patients (excluding those with medical education) were randomized to review one version.
• Readability assessed using Flesch Reading Ease (FRE) and LIX scores; patient comprehension and engagement evaluated via questionnaire.

Main Results:

• Simplified reports (texts 2 and 3) demonstrated significantly higher readability scores (FRE) and lower LIX scores compared to the original (text 1).
• Text 3, the most simplified version, achieved the best scores in patient-rated clarity, tone, structure, and engagement.
• Patients reported improved ability to understand findings without assistance and felt better prepared for clinical conversations with simplified reports.

Conclusions:

• AI-driven simplification of radiology reports substantially improves patient comprehension and engagement.
• ChatGPT shows promise as a tool for enhancing patient-centered communication in healthcare.
• Further research is needed to ensure clinical accuracy and explore equitable integration across diverse patient populations.