Standardized evaluation of diabetic retinopathy using artificial intelligence and its association with metabolic dysfunction-associated steatotic liver disease in Japan: A cross-sectional study
2024

AI Evaluation of Diabetic Retinopathy and Liver Disease in Japan

Sample size: 1736 publication 10 minutes Evidence: moderate

Author Information

Author(s): Komatsu Koji, Sano Kei, Fukai Kota, Nakagawa Ryo, Nakagawa Takashi, Tatemichi Masayuki, Nakano Tadashi

Primary Institution: Jikei University School of Medicine, Tokyo, Japan

Hypothesis

This study aimed to evaluate fundus photographs using artificial intelligence to explore the relationships between diabetic retinopathy (DR), metabolic dysfunction-associated steatotic liver disease (MASLD), and related factors.

Conclusion

The study demonstrates that MASLD is associated with a reduced risk of diabetic retinopathy, with higher cholinesterase levels providing further protective effects.

Supporting Evidence

  • MASLD affects 30% of the general adult population and 60-70% of patients with diabetes and obesity.
  • Diabetic retinopathy is a common complication of diabetes and has been linked to liver disease.
  • AI software RetCAD achieved a 95.1% area under the ROC curve for detecting diabetic retinopathy.

Takeaway

This study used AI to look at eye pictures from people with diabetes and found that those with liver disease might have a lower chance of eye problems.

Methodology

This cross-sectional study analyzed fundus photographs of 1,736 patients with diabetes using AI software to evaluate diabetic retinopathy and assessed liver disease stages through ultrasound.

Potential Biases

Potential misclassification of diabetic retinopathy due to varying experience levels of ophthalmologists.

Limitations

The study is cross-sectional, making it difficult to establish causality, and it may have omitted peripheral retinal changes due to the use of posterior-pole fundus images.

Participant Demographics

Participants were aged 35-75 years, with a male predominance (77.9%).

Statistical Information

P-Value

0.02

Confidence Interval

0.43–0.95

Statistical Significance

p<0.05

Digital Object Identifier (DOI)

10.1371/journal.pone.0315752

Want to read the original?

Access the complete publication on the publisher's website

View Original Publication