Study demonstrates the potential of using artificial intelligence in oculomics to assess HbA1c levels from fundus images. By analyzing over 6,000 images, the researchers found that diverse datasets are key to minimizing bias and improving model reliability across different populations.