Synapse
⌘+K
Synapse
PulseExploreClubsResearchersJournals
Instagram
HomeClubsExplore
September 10, 2025European Journal of TherapeuticsOpen Access

Evaluation of the Performance of ChatGPT 4.5 in LI-RADS Categorization and Management Suggestion: Zero-shot versus Few-shot Prompting Method

View Full Paper
Ask AI
Bookmark
Share

Authors

EÇEren ÇamurYGYasin Celal Güneş

Discussion

Loading...

Member takes

Overview

Cross-sectional study assesses chatgpt 4.5 using few-shot prompting, indicating improved management suggestions and classification accuracy.

Key Points

  • ChatGPT 4.5 produced 92% correct LI-RADS classifications with few-shot prompting, outperforming the radiologist's 82%.
  • The model achieved 84% accuracy for management recommendations under few-shot prompting, significantly enhancing clinical utility.
  • Statistical analysis indicated significant improvement in management suggestions (p=0.016) compared to zero-shot prompting.
  • This early benchmark study highlights the potential of chatgpt 4.5 in advancing diagnostic tools for radiologists.

Cite This Study

Çamur et al. (2025) studied this question.

synapsesocial.com/papers/68c23c66b210217d64788be4https://doi.org/10.58600/eurjther2699
View Full Paper
Ask AI
Bookmark
Share