Artificial Intelligence Software for Diabetic Eye Screening: Diagnostic Performance and Impact of Stratification

Abstract

Aim: To evaluate the MONA.health artificial intelligence screening software for detecting referable diabetic retinopathy (DR) and diabetic macular edema (DME), including subgroup analysis. Methods: The algorithm’s threshold value was fixed at the 90% sensitivity operating point on the receiver operating curve to perform the disease classification. Diagnostic performance was appraised on a private test set and publicly available datasets. Stratification analysis was executed on the private test set considering age, ethnicity, sex, insulin dependency, year of examination, camera type, image quality, and dilatation status. Results: The software displayed an area under the curve (AUC) of 97.28% for DR and 98.08% for DME on the private test set. The specificity and sensitivity for combined DR and DME predictions were 94.24 and 90.91%, respectively. The AUC ranged from 96.91 to 97.99% on the publicly available datasets for DR. AUC values were above 95% in all subgroups, with lower predictive values found for individuals above the age of 65 (82.51% sensitivity) and Caucasians (84.03% sensitivity). Conclusion: We report good overall performance of the MONA.health screening software for DR and DME. The software performance remains stable with no significant deterioration of the deep learning models in any studied strata.

Diabetic Macular Edema Optical Coherence Tomography Biomarkers Detected with EfficientNetV2B1 and ConvNeXt

“…In addition, the sensitivity ranges from 87 to 100% and specificity is 73.3–98.5% for DR, making it more suitable for screening [ 15 ]. Other authors developed similar AI models capable of predicting DR and DME with high accuracy on fundus images [ 16 ].…”

“…The most important disadvantages of the mainstream screening programs worldwide are the lack of modern equipment, small number of ophthalmologists in relation to the population size, and high healthcare costs for screening large populations [ 16 , 17 , 18 , 19 ].…”

Insights into the liver-eyes connections, from epidemiological, mechanical studies to clinical translation

“…In addition, artificial intelligence (AI)-based deep-learning (DL) algorithm has evolved rapidly in medical imaging-processing by automatically analyzing input graphs and coming to diagnosis data [ 174 , 175 ]. This technology has been validated in image-centered ophthalmology to detect glaucoma, multiple retinopathies (including ROP, AMD, DR, and diabetic macular edema) [ 176 – 180 ]. The latest study adopts the ResNet-101 deep neural network to establish both slit-lamp and fundus images-trained DL models, in which the slit-lamp model performs well in detecting liver cirrhosis and cancer, while both models work relatively weaker in predicting cholelithiasis, NAFLD, viral hepatitis, and hepatic cysts [ 2 ].…”