bims-aukdir Biomed News
on Automated knowledge discovery in diabetes research
Issue of 2025–08–24
eight papers selected by
Mott Given
  1. Real-World Evaluation of Artificial Intelligence-Based Diabetic Retinopathy Screening Using the Optomed Aurora Handheld Fundus Camera.
  2. Artificial Intelligence Improves Patient Follow-Up in a Diabetic Retinopathy Screening Program [Letter].
  3. Comparative performance of deep learning architectures for diabetic peripheral neuropathy detection using corneal confocal microscopy: a retrospective single-centre study.
  4. Deep learning for retinal non-perfusion and foveal avascular zone analysis in wide-field OCTA in diabetic retinopathy.
  5. Human expert grading versus automated quantification of fluid volumes in nAMD, DME and BRVO.
  6. Global Research Trends in Artificial Intelligence and Type 2 Diabetes Mellitus: A Bibliometric Perspective.
  7. Correlation study on the risk of cardiovascular adverse events in diabetic foot patients based on machine learning - a retrospective cohort study.
  8. Deep learning applications for diabetic retinopathy and retinopathy of prematurity diseases diagnosis: a systematic review.
  1. Diabetes Technol Ther. 2025 Aug 18.
    Real-World Evaluation of Artificial Intelligence-Based Diabetic Retinopathy Screening Using the Optomed Aurora Handheld Fundus Camera.
    Petri Huhtinen, Anna-Maria Kubin, Kamila Dvořák, Martin Sliva, Jan Bayer, Nina Hautala.
      Diabetic retinopathy (DR) is a common and potentially sight-threatening complication of diabetes. Early detection of DR through screening can prevent visual loss. Handheld fundus cameras combined with artificial intelligence (AI) technology may improve DR screening. We evaluated the Aireen AI algorithm's performance in grading DR in fundus images captured by the handheld Optomed Aurora. Two retina specialists and Aireen graded 624 fundus images for DR. Sensitivity, specificity, and predictive values were measured against the ophthalmologists' grading. Overall, 97% of images were sufficient for DR classification. Aireen demonstrated 94.8% sensitivity, 91.4% specificity, and 92.7% diagnostic accuracy for DR. Aireen showed high diagnostic accuracy in detecting DR in Optomed Aurora images, suggesting its potential for effective screening. The validated use of AI with a handheld fundus camera may streamline the screening process, reduce the burden on health care professionals, and improve access to screening and patient outcomes through enhanced diagnostic accuracy.
    Keywords:  artificial intelligence; diabetic retinopathy; handheld fundus camera; screening for diabetic retinopathy
    DOI:  https://doi.org/10.1177/15209156251369886
  2. Clin Ophthalmol. 2025 ;19 2659-2660
    Artificial Intelligence Improves Patient Follow-Up in a Diabetic Retinopathy Screening Program [Letter].
    Avi P Bhavsar, Charles I Minerd, Niloofar Piri.
      
    DOI:  https://doi.org/10.2147/OPTH.S537960
  3. BMJ Open. 2025 Aug 19. 15(8): e095342
    Comparative performance of deep learning architectures for diabetic peripheral neuropathy detection using corneal confocal microscopy: a retrospective single-centre study.
    Wenqu Chen, Yuyang Deng, Weihuang Xu, Jianzhang Hu.
       OBJECTIVES: This study aims to develop a deep learning algorithm (DLA) using the InceptionV3 architecture for effective diabetic peripheral neuropathy (DPN) screening via corneal confocal microscopy (CCM) images.
    DESIGN: Retrospective study.
    SETTING: Ophthalmology Centre of General Hospital.
    PARTICIPANTS: 127 participants enrolled: 33 healthy participants, 57 diabetic patients with DPN (DPN+) and 37 diabetic patients without DPN (DPN-).
    INTERVENTIONS: Not applicable.
    MAIN OUTCOME MEASURES: The CCM image dataset, which was collected from participants (with five images per eye), was randomly divided into training, validation and test subsets in a 7:1:2 ratio. The images were preprocessed, augmented and used to train the InceptionV3 model. We compared its performance against the ResNet, DenseNet and Swin Transformer models. Performance was evaluated using accuracy, recall, F1 score and area under the curve (AUC) metrics.
    RESULTS: For single-participant predictions, the InceptionV3 model achieved the highest accuracy (0.9231), recall (0.8846), F1 score (0.9020) and AUC (0.9534) compared with the other models. For single-image predictions in the three-class classification task of CCM images, the InceptionV3 model achieved a precision of 0.8385, a recall of 0.9083, an F1 score of 0.8720 and an AUC of 0.8769 for predicting DPN+.
    CONCLUSIONS: The InceptionV3-based DLA model achieved superior performance compared with traditional convolutional neural network architectures like ResNet and DenseNet, and the Swin transformer model, highlighting its potential for effective DPN screening.
    Keywords:  Artificial Intelligence; Corneal and external diseases; Diabetes Mellitus, Type 2; Diabetic neuropathy; Machine Learning
    DOI:  https://doi.org/10.1136/bmjopen-2024-095342
  4. Sci Rep. 2025 Aug 18. 15(1): 30225
    Deep learning for retinal non-perfusion and foveal avascular zone analysis in wide-field OCTA in diabetic retinopathy.
    Hugo Le Boité, Sophie Bonnin, Mathias Gallardo, Mathieu Lamard, Aude Couturier, Gwenolé Quellec.
      We developed an automated framework for segmenting low-quality and non-perfusion areas in widefield OCTA images to obtain two key metrics useful for diabetic retinopathy (DR) monitoring: the retinal non-perfusion index (NPI) and foveal avascular zone (FAZ) area. Using 170 images from 88 patients in the EVIRED cohort, we trained two models: Q-NET, which segments low-quality areas, and NPA-NET, which detects non-perfusion areas and the FAZ. Their combined outputs created a 4-class map to calculate NPI and FAZ area. Ground truth segmentations were established by a single expert (for non-perfusion and FAZ areas) or a consensus of four annotators (for low-quality areas). NPA-NET and Q-NET, tested on 29 images, achieved strong segmentation performances (Dice coefficients of 0.714 (low-quality), 0.781 (non-perfusion), and 0.879 (FAZ)). Some inter-annotator variability was found (mean Dice: 0.85 for low-quality, 0.683 for non-perfusion areas). Predictive accuracy for NPI and FAZ area was high, with R² coefficients of 0.97 and 0.63, respectively, with minimal underestimation and no overestimation. This AI tool provides reliable biomarkers for DR monitoring, supporting treatment decisions and medical decision-making by automatically analyzing OCTA images, and could be integrated into clinical practice.
    Keywords:  Artificial intelligence; Diabetic retinopathy; Image segmentation; OCT-angiography; Retinal non-perfusion
    DOI:  https://doi.org/10.1038/s41598-025-15712-3
  5. Sci Rep. 2025 Aug 19. 15(1): 30414
    Human expert grading versus automated quantification of fluid volumes in nAMD, DME and BRVO.
    Felix Goldbach, Bianca S Gerendas, Oliver Leingang, Thomas Alten, Alexandros Bampoulidis, Jonas Brugger, Hrvoje Bogunovic, Amir Sadeghipour, Ursula Schmidt-Erfurth.
      This study compared an automated deep learning algorithm with certified human graders from the Vienna Reading Center (VRC) in identifying intra- (IRF) and subretinal fluid (SRF) in OCT scans of patients treated for neovascular age-related macular degeneration (nAMD), diabetic macular edema (DME) and branch retinal vein occlusion (BRVO). Multicenter clinical trial data from the VRC imaging database was used for this post hoc analysis. OCT scans were analyzed using a validated algorithm (RetInSight, Vienna, Austria) to compute IRF and SRF volumes. These fluid volumes were compared to fluid presence graded by trained and experienced graders of the VRC. 6898 OCT scans were analyzed for fluid volumes and presence of IRF and SRF. For nAMD/DME /BRVO in the central millimeter: the overall concordance for the detection of IRF and SRF between the algorithm and manual grading reached an AUC of 0.94/0.92/0.98 and 0.89/0.95/0.92, respectively. This deep learning approach showed a high concordance with human expert grading for detection of IRF and SRF and provides precise volumetric information across different retinal fluid-associated diseases. Thus, automated fluid quantification is a feasible tool for standardized treatment decision support and disease monitoring in clinical practice at the highest human expert level.
    DOI:  https://doi.org/10.1038/s41598-025-13019-x
  6. Cureus. 2025 Jul;17(7): e88114
    Global Research Trends in Artificial Intelligence and Type 2 Diabetes Mellitus: A Bibliometric Perspective.
    Paul Anthony Camacho López, Maria G Latorre-Arevalo, Pablo Camacho-Naranjo, Silvia J Villabona-Florez.
      Artificial intelligence (AI) applied to type 2 diabetes mellitus (T2DM) is transforming the diagnosis and management of this chronic disease, posing a significant public health challenge. Despite recent advances, there remains a gap in the systematization of knowledge regarding AI and T2DM, as well as in the identification of trends and scientific collaborations in this field. This study aims to conduct a comprehensive bibliometric analysis of academic output on AI applied to T2DM, mapping the main actors, collaboration networks, and predominant research themes from 2000 to 2024. A bibliometric analysis was conducted using the Web of Science database, focusing on scientific topics related to AI applied to T2DM from 2000 to 2024. Bibliometric tools such as Bibliometrix, VOSviewer, and CitNetExplorer were utilized to examine publication patterns, co-citation networks, and keywords. The analysis included 1,454 original articles and 134 reviews, aiming to identify the most influential authors, institutions, and countries in the field. The analysis revealed a growth rate of 1.7%, with significant increases observed between 2020 and 2024. The research highlighted the use of AI for the detection of diabetic retinopathy and continuous glucose monitoring as the primary areas of publication. China (27.4%) and India (20.5%) lead scientific production and international collaborations in this field, reflecting the globalization of health research. This study provides an overview of the current state and future opportunities in AI research applied to T2DM. The findings are valuable for researchers, healthcare professionals, and academic institutions, fostering progress in AI and T2DM through collaborative and ethical strategies. This bibliometric analysis contributes to guiding the development of health research policies and optimizing the use of AI in managing T2DM.
    Keywords:  artificial intelligence; bibliometrics; bibliometrix; citnetexplorer; machine learning; type 2 diabetes mellitus; vosviewer
    DOI:  https://doi.org/10.7759/cureus.88114
  7. Front Endocrinol (Lausanne). 2025 ;16 1595471
    Correlation study on the risk of cardiovascular adverse events in diabetic foot patients based on machine learning - a retrospective cohort study.
    Liran Zheng, Jiageng Chen, Wenyan Xu, Min Ding, Juan Li, Fenghua Tian, Lei Zhang, Qianqian Li, Shuai Wang, Zeyu Wang, Hairong Ma, Xuecan Cui, Bai Chang, Meijun Wang.
       Introduction: Diabetic Foot (DF), as a serious complication of diabetes, is closely related to major adverse cardiovascular events (MACE) and mortality. However, research on predictive models for the MACE risk in DF patients is not sufficient. The purpose of this study is to construct a prognostic model for the MACE risk in patients with diabetic foot ulcers and provide a reference tool for clinical individualized management.
    Method: This study retrospectively collected data of DF patients who were hospitalized and met the inclusion and exclusion criteria in a tertiary first-class comprehensive hospital mainly engaged in metabolic diseases in Tianjin from January 2018 to January 2020. The follow-up outcome was the occurrence of MACE within 5 years after discharge. Multiple imputation (MI) method was used to fill in the missing data. Based on the processed data, in terms of modeling methods, the top three frequently used methods were used. Logistic regression, random forest (RF) and support vector machine (SVM) were used respectively to analyze influencing factors. The performance of each model was compared by using confusion matrix, ROC curve and AUC value. The data set was divided into training set and test set according to the proportion of 80%/20%. Finally, the model effect was verified on the test set. The study finally included a total of 504 patients with DF. Among them, 147 cases (29.17%) experienced MACE events within five years. The AUC of the RF model in this study was 0.70, the AUC of the Logistic regression model was 0.62, and the AUC of the SVM model was 0.60.
    Conclusion: All three models established in this research have good clinical predictive ability. Among them, the clinical prediction model based on RF has the best effect and can effectively predict the risk of MACE in DF patients, helping clinical medical staff formulate personalized treatment plans.
    Keywords:  diabetic foot; major adverse cardiovascular events; random forest mode; relevance; risk prediction
    DOI:  https://doi.org/10.3389/fendo.2025.1595471
  8. Int J Ophthalmol. 2025 ;18(8): 1594-1602
    Deep learning applications for diabetic retinopathy and retinopathy of prematurity diseases diagnosis: a systematic review.
    Elizabeth Ndunge Mutua, Bernard Shibwabo Kasamani, Christoph Reich.
      To review the existing deep learning applications for diagnosing diabetic retinopathy and retinopathy of prematurity diseases, the available public retinal databases for the diseases and apply the International Journal of Medical Informatics (IJMEDI) checklist were assessed the quality of included studies; an in-depth literature search in Scopus, Web of Science, IEEE and ACM databases targeting articles from inception up to 31st January 2023 was done by two independent reviewers. In the review, 26 out of 1476 articles with a total of 36 models were included. Data size and model validation were found to be challenges for most studies. Deep learning models are gaining focus in the development of medical diagnosis tools and applications. However, there seems to be a critical issue with most of the studies being published, with some not including information about data sources and data sizes which is important for their performance verification.
    Keywords:  deep learning; diabetic retinopathy; retinal database; retinal vessel segmentation; retinopathy of prematurity
    DOI:  https://doi.org/10.18240/ijo.2025.08.23
This page is being maintained by Thomas Krichel. It was last updated on 2025‒08‒24 at 0:14.