An inherently interpretable AI model improves screening speed and accuracy for early diabetic retinopathy
PLOS Digital Health · 12.05.2025 · 4 (5) · e0000831
DOI: 10.1371/journal.pdig.0000831 ↗Kurz gefasst
Fachartikel in der Zeitschrift PLOS Digital Health, online erschienen am 12.05.2025. Verfasst von Kerol Djoumessi und 6 weiteren Autor:innen. Beteiligte Einrichtung: Hertie Institute for AI in Brain Health, University of Tübingen, Tübingen, Germany. Die Arbeit wurde bisher 13-mal zitiert (Crossref) und nennt 49 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Deutsche Forschungsgemeinschaft, Gemeinnützige Hertie-Stiftung, Carl-Zeiss-Stiftung.
Abstract (Originalsprache)
Diabetic retinopathy (DR) is a frequent complication of diabetes, affecting millions worldwide. Screening for this disease based on fundus images has been one of the first successful use cases for modern artificial intelligence in medicine. However, current state-of-the-art systems typically use black-box models to make referral decisions, requiring post-hoc methods for AI-human interaction and clinical decision support. We developed and evaluated an inherently interpretable deep learning model, which explicitly models the local evidence of DR as part of its network architecture, for clinical decision support in early DR screening. We trained the network on 34,350 high-quality fundus images from a publicly available dataset and validated its performance on a large range of ten external datasets. The inherently interpretable model was compared to post-hoc explainability techniques applied to a standard DNN architecture. For comparison, we obtained detailed lesion annotations from ophthalmologists on 65 images to study if the class evidence maps highlight clinically relevant information. We tested the clinical usefulness of our model in a retrospective reader study, where we compared screening for DR without AI support to screening with AI support with and without AI explanations. The inherently interpretable deep learning model obtained an accuracy of .906 [.900–.913] (95%-confidence interval) and an AUC of .904 [.894–.913] on the internal test set and similar performance on external datasets, comparable to the standard DNN. High evidence regions directly extracted from the model contained clinically relevant lesions such as microaneurysms or hemorrhages with a high precision of .960 [.941–.976], surpassing post-hoc techniques applied to a standard DNN. Decision support by the model highlighting high-evidence regions in the image improved screening accuracy for difficult decisions and improved screening speed. This shows that inherently interpretable deep learning models can provide clinical decision support while obtaining state-of-the-art performance improving human-AI collaboration.
Autor:innen und Institutionen
- Hertie Institute for AI in Brain Health, University of Tübingen, Tübingen, Germany.
- Tübingen AI Center, University of Tübingen, Tübingen, Germany.
- University Eye Hospital, University of Tübingen, Tübingen, Germany.
- Eye Clinic Sulzbach, Knappschaft Hospital Saar, Sulzbach, Germany.
- Black Forest Eye Clinic, Endingen, Germany.
- Department of Diabetes, Endocrinology, Nutritional Medicine and Metabolism UDEM, Inselspital, Bern University Hospital, University of Bern, Bern, Switzerland.
ORCID-Profile: Kerol Djoumessi, Laura Kühlewein, Philipp Berens
Förderung
- Deutsche Forschungsgemeinschaft
- Gemeinnützige Hertie-Stiftung
- Carl-Zeiss-Stiftung
Zitieren
APA
Djoumessi, K., Huang, Z., Kühlewein, L., Rickmann, A., Simon, N., Koch, L. M., & Berens, P. (2025). An inherently interpretable AI model improves screening speed and accuracy for early diabetic retinopathy. PLOS Digital Health, 4(5), e0000831. https://doi.org/10.1371/journal.pdig.0000831
Vancouver
Djoumessi K, Huang Z, Kühlewein L, Rickmann A, Simon N, Koch LM, et al. An inherently interpretable AI model improves screening speed and accuracy for early diabetic retinopathy. PLOS Digital Health. 2025;4(5):e0000831. doi:10.1371/journal.pdig.0000831
BibTeX
@article{Djoumessi2025,
author = {Djoumessi, Kerol and Huang, Ziwei and Kühlewein, Laura and Rickmann, Annekatrin and Simon, Natalia and Koch, Lisa M and Berens, Philipp},
title = {An inherently interpretable AI model improves screening speed and accuracy for early diabetic retinopathy},
journal = {PLOS Digital Health},
year = {2025},
volume = {4},
number = {5},
pages = {e0000831},
doi = {10.1371/journal.pdig.0000831}
}
EUVIMED