PRECISE framework: Enhanced radiology reporting with GPT for improved readability, reliability, and patient-centered...

Area of Science:

• Medical Imaging Informatics
• Artificial Intelligence in Healthcare
• Patient Communication

Background:

• The PRECISE framework (Patient-Focused Radiology Reports with Enhanced Clarity and Informative Summaries for Effective Communication) utilizes GPT-4.
• It aims to generate patient-friendly summaries of radiology reports at a sixth-grade reading level.

Purpose of the Study:

• To evaluate the effectiveness of the PRECISE framework in improving radiology report readability, reliability, and understandability.
• To test the hypothesis that PRECISE enhances report readability and patient comprehension compared to original reports.

Main Methods:

• Assessed the PRECISE framework on 500 chest X-ray reports.
• Evaluated readability using Flesch Reading Ease, Gunning Fog Index, and Automated Readability Index.
• Gauged reliability with clinical volunteers and understandability with non-medical volunteers, using statistical analyses.

Main Results:

• Significant improvements in readability scores were observed (e.g., Flesch Reading Ease increased from 38.28 to 80.82, p < 0.001).
• 95% of summaries were deemed reliable by clinical volunteers.
• 97% of PRECISE-generated summaries were rated fully understandable by non-medical volunteers.

Conclusions:

• The PRECISE framework shows promise for enhancing patient understanding and communication without increasing radiologist burden.
• Improved reliability and patient-friendly summaries can foster patient engagement in healthcare decisions.
• This AI approach marks a significant step toward more inclusive, patient-centric care delivery.