This study presents the development and validation of a deep learning system, DeepDR Plus, designed to predict the time to progression of diabetic retinopathy (DR) within 5 years using fundus images alone. The system was pre-trained on 717,308 fundus images from 179,327 participants with diabetes and then validated on a multiethnic dataset of 118,868 images from 29,868 participants. The DeepDR Plus system achieved concordance indexes of 0.754–0.846 and integrated Brier scores of 0.153–0.241 for all times up to 5 years. In real-world cohorts, the system could potentially extend the mean screening interval from 12 months to 31.97 months, with a delayed detection rate of only 0.18%. The system also demonstrated good performance in predicting DR progression in three subgroups: from no retinopathy to DR, from non-referable DR to referable DR, and from non-vision-threatening DR to vision-threatening DR. The integration of the DeepDR Plus system into clinical workflows showed significant improvements in patient outcomes, with a reduced rate of DR progression and a lower mean screening frequency. The study highlights the potential of AI-driven personalized screening intervals to enhance the efficiency and equity of DR management.This study presents the development and validation of a deep learning system, DeepDR Plus, designed to predict the time to progression of diabetic retinopathy (DR) within 5 years using fundus images alone. The system was pre-trained on 717,308 fundus images from 179,327 participants with diabetes and then validated on a multiethnic dataset of 118,868 images from 29,868 participants. The DeepDR Plus system achieved concordance indexes of 0.754–0.846 and integrated Brier scores of 0.153–0.241 for all times up to 5 years. In real-world cohorts, the system could potentially extend the mean screening interval from 12 months to 31.97 months, with a delayed detection rate of only 0.18%. The system also demonstrated good performance in predicting DR progression in three subgroups: from no retinopathy to DR, from non-referable DR to referable DR, and from non-vision-threatening DR to vision-threatening DR. The integration of the DeepDR Plus system into clinical workflows showed significant improvements in patient outcomes, with a reduced rate of DR progression and a lower mean screening frequency. The study highlights the potential of AI-driven personalized screening intervals to enhance the efficiency and equity of DR management.