Impact of AI on Polyp Size and Colonoscopy
Author Information
Author(s): Muacevic Alexander, Adler John R, Yousaf Muhammad N, Sharma Neal, Matteson-Kome Michelle L, Puli Srinivas, Nguyen Douglas, Bechtold Matthew L
Primary Institution: University of Missouri School of Medicine
Hypothesis
Can artificial intelligence improve the accuracy of polyp size estimation compared to experienced gastroenterologists?
Conclusion
AI appears to be more accurate in estimating polyp size than experienced endoscopists, potentially leading to better surveillance intervals and cost savings.
Supporting Evidence
- AI had a mean accuracy rate of 96% compared to 76% for gastroenterologists.
- Endoscopists estimated polyp size within ± 1 mm 36% of the time, while AI did so 82% of the time.
- Surveillance recommendations by endoscopists were inappropriate 34 times compared to 0 times for AI.
Takeaway
This study shows that AI can help doctors measure the size of polyps in the colon more accurately than they can do by themselves, which can help in deciding when to check patients again.
Methodology
A survey study using a phantom colon model where gastroenterologists' polyp size estimations were compared to an AI system.
Potential Biases
Potential bias due to the subjective nature of visual estimation by gastroenterologists.
Limitations
The study had a small sample size and limited baseline data on polyp size differences.
Participant Demographics
Mean age of gastroenterologists was 51.1 years with an average of 19.3 years of experience.
Statistical Information
P-Value
p<0.01
Statistical Significance
p<0.01
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website