bims-aukdir Biomed News
on Automated knowledge discovery in diabetes research
Issue of 2025–05–18
twelve papers selected by
Mott Given



  1. Sci Rep. 2025 May 12. 15(1): 16393
      Macrovascular complications are leading causes of morbidity and mortality in patients with type 2 diabetes mellitus (T2DM), yet early diagnosis of cardiovascular disease (CVD) in this population remains clinically challenging. This study aims to develop a machine learning model that can accurately predict diabetic macroangiopathy in Chinese patients. A retrospective cross-sectional analytical study was conducted on 1566 hospitalized patients with T2DM. Feature selection was performed using recursive feature elimination (RFE) within the mlr3 framework. Model performance was benchmarked using 29 machine learning (ML) models, with the ranger model selected for its superior performance. Hyperparameters were optimized through grid search and 5-fold cross-validation. Model interpretability was enhanced using SHAP values and PDPs. An external validation set of 106 patients was used to test the model. Key predictive variables identified included the duration of T2DM, age, fibrinogen, and serum urea nitrogen. The predictive model for macroangiopathy was established and showed good discrimination performance with an accuracy of 0.716 and an AUC of 0.777 in the training set. Validation on the external dataset confirmed its robustness with an AUC of 0.745. This study establish an approach based on machine learning algorithm in features selection and the development of prediction tools for diabetic macroangiopathy.
    Keywords:  Machine learning methods; Macroangiopathy; Prediction model; Risk factor; T2DM
    DOI:  https://doi.org/10.1038/s41598-025-01161-5
  2. PLOS Digit Health. 2025 May;4(5): e0000831
      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.
    DOI:  https://doi.org/10.1371/journal.pdig.0000831
  3. Conf Comput Vis Pattern Recognit Workshops. 2024 Jun;2024 2285-2294
      Over the past few decades, convolutional neural networks (CNNs) have been at the forefront of the detection and tracking of various retinal diseases (RD). Despite their success, the emergence of vision transformers (ViT) in the 2020s has shifted the trajectory of RD model development. The leading-edge performance of ViT-based models in RD can be largely credited to their scalability-their ability to improve as more parameters are added. As a result, ViT-based models tend to outshine traditional CNNs in RD applications, albeit at the cost of increased data and computational demands. ViTs also differ from CNNs in their approach to processing images, working with patches rather than local regions, which can complicate the precise localization of small, variably presented lesions in RD. In our study, we revisited and updated the architecture of a CNN model, specifically MobileNet, to enhance its utility in RD diagnostics. We found that an optimized MobileNet, through selective modifications, can surpass ViT-based models in various RD benchmarks, including diabetic retinopathy grading, detection of multiple fundus diseases, and classification of diabetic macular edema. The code is available at https://github.com/Retinal-Research/NN-MOBILENET.
    DOI:  https://doi.org/10.1109/CVPRW63382.2024.00234
  4. PLOS Digit Health. 2025 May;4(5): e0000853
      Counterfactual reasoning is often used by humans in clinical settings. For imaging based specialties such as ophthalmology, it would be beneficial to have an AI model that can create counterfactual images, illustrating answers to questions like "If the subject had had diabetic retinopathy, how would the fundus image have looked?". Such an AI model could aid in training of clinicians or in patient education through visuals that answer counterfactual queries. We used large-scale retinal image datasets containing color fundus photography (CFP) and optical coherence tomography (OCT) images to train ordinary and adversarially robust classifiers that classify healthy and disease categories. In addition, we trained an unconditional diffusion model to generate diverse retinal images including ones with disease lesions. During sampling, we then combined the diffusion model with classifier guidance to achieve realistic and meaningful counterfactual images maintaining the subject's retinal image structure. We found that our method generated counterfactuals by introducing or removing the necessary disease-related features. We conducted an expert study to validate that generated counterfactuals are realistic and clinically meaningful. Generated color fundus images were indistinguishable from real images and were shown to contain clinically meaningful lesions. Generated OCT images appeared realistic, but could be identified by experts with higher than chance probability. This shows that combining diffusion models with classifier guidance can achieve realistic and meaningful counterfactuals even for high-resolution medical images such as CFP images. Such images could be used for patient education or training of medical professionals.
    DOI:  https://doi.org/10.1371/journal.pdig.0000853
  5. Prog Retin Eye Res. 2025 May 11. pii: S1350-9462(25)00036-9. [Epub ahead of print] 101363
      The adoption of standardized imaging protocols in retinal imaging is critical to overcoming challenges posed by fragmented data formats across devices and manufacturers. The lack of standardization hinders clinical interoperability, collaborative research, and the development of artificial intelligence (AI) models that depend on large, high-quality datasets. The Digital Imaging and Communication in Medicine (DICOM) standard offers a robust solution for ensuring interoperability in medical imaging. Although DICOM is widely utilized in radiology and cardiology, its adoption in ophthalmology remains limited. Retinal imaging modalities such as optical coherence tomography (OCT), fundus photography, and OCT angiography (OCTA) have revolutionized retinal disease management but are constrained by proprietary and non-standardized formats. This review underscores the necessity for harmonized imaging standards in ophthalmology, detailing DICOM standards for retinal imaging including ophthalmic photography (OP), OCT, and OCTA, and their requisite metadata information. Additionally, the potential of DICOM standardization for advancing AI applications in ophthalmology is explored. A notable example is the Artificial Intelligence Ready and Equitable Atlas for Diabetes Insights (AI-READI) dataset, the first publicly available standards-compliant DICOM retinal imaging dataset. This dataset encompasses diverse retinal imaging modalities, including color fundus photography, infrared, autofluorescence, OCT, and OCTA. By leveraging multimodal retinal imaging, AI-READI provides a transformative resource for studying diabetes and its complications, setting a blueprint for future datasets aimed at harmonizing imaging formats and enabling AI-driven breakthroughs in ophthalmology. Our manuscript also addresses challenges in retinal imaging for diabetic patients, retinal imaging-based AI applications for studying diabetes, and potential advancements in retinal imaging standardization.
    Keywords:  Arterial intelligence; Artificial Intelligence Ready and Equitable Atlas for Diabetes Insights; DICOM; Data standardization; Diabetic retinopathy; Imaging; Interoperability
    DOI:  https://doi.org/10.1016/j.preteyeres.2025.101363
  6. J Diabetes Metab Disord. 2025 Jun;24(1): 115
       Background and objectives: Social determinants of health (SDOH) play a critical role in the onset and progression of chronic kidney disease (CKD). Despite the well-established role of SDOH, previous studies have not fully incorporated these factors in predicting CKD in Type 2 diabetes patients. To bridge this gap, this study aimed to develop and evaluate the machine learning (ML) models that incorporate SDOH to enhance CKD risk prediction in Type 2 diabetes patients.
    Methods: Data were obtained from the 2023 Behavioral Risk Factor Surveillance System (BRFSS), a national survey that collects comprehensive health-related data from adults across the United States. Missing data were addressed using the K-nearest neighbor imputation method, and the Synthetic Minority Oversampling Technique (SMOTE) was applied to balance class distributions. Potential predictive features were selected using correlation coefficient analysis. The dataset was partitioned into training (80%) and testing (20%) subsets, with a 3-fold cross-validation strategy applied to the training data. Seven ML models were developed for CKD risk prediction, including logistic regression (LR), decision tree (DT), K-nearest neighbor (KNN), random forest (RF), eXtreme Gradient Boosting (XGBoost), and an artificial neural network (ANN). Model performance was evaluated using multiple metrics, including the area under the receiver operating characteristic curve (AUROC), precision, recall, F1 score, accuracy, and false positive rate.
    Results: The study included 19,912 Type 2 diabetes patients (weighted sample size: 818,878), among whom 2,924 (weighted 13.92%) had CKD, and 16,988 (weighted 86.08%) did not. Over half of the CKD group (50.4%) were aged 65 or older. The proportion of female patients was higher in both groups, comprising 53.8% of the CKD group and 50.5% of the non-CKD group. Among the ML models evaluated, the RF model demonstrated the highest predictive performance for CKD, with an AUROC of 0.89 (95% CI: 0.88 - 0.90), followed by the DT model (0.84, 95% CI: 0.83 - 0.85) and XGBoost (0.83, 95% CI: 0.82 - 0.84). The RF model achieved an accuracy of 0.81 (95%CI: 0.81 - 0.81), a precision of 0.79 (95%CI: 0.79 - 0.79), a recall of 0.85 (95%CI: 0.85 - 0.85), and an F1 score of 0.82 (95%CI: 0.82 - 0.82). Additionally, the RF model exhibited strong calibration, reinforcing its reliability as a predictive tool for CKD risk in individuals with Type 2 diabetes.
    Conclusion: The study findings underscore the potential of ML models, particularly the RF model, in accurately predicting CKD among individuals with Type 2 diabetes. This approach not only enhances the precision of CKD prediction but also highlights the importance of addressing social and environmental disparities in disease prevention and management. Leveraging ML models with SDOH can lead to earlier interventions, more personalized treatment plans, and improved health outcomes for vulnerable populations.
    Supplementary Information: The online version contains supplementary material available at 10.1007/s40200-025-01621-9.
    Keywords:  Chronic kidney disease; Machine learning; Random forest; Social determinants of health; Type 2 diabetes; XGBoost
    DOI:  https://doi.org/10.1007/s40200-025-01621-9
  7. Artif Intell Med. 2025 May 02. pii: S0933-3657(25)00088-0. [Epub ahead of print]166 103153
      Chronic kidney disease (CKD) poses a significant risk for diabetes patients, often leading to severe complications. Early and accurate CKD stage detection is crucial for timely intervention. However, it remains challenging due to its asymptomatic progression, the oversight of routine CKD tests during diabetes checkups, and limited access to nephrologists. This study aimed to address these challenges by developing a multiclass CKD stage prediction model for diabetes patients using longitudinal data from the Chronic Renal Insufficiency Cohort (CRIC) study. A novel iterative backward feature selection strategy was employed to determine key predictors of the CKD stage. TabNet, an attention-based deep learning architecture, was used to build classification models in complete and simplified categories. The complete model used 31 features, including complex kidney biomarkers, while the simplified model used 15 features readily available from routine checkups. The performance of TabNet was compared against traditional tree-based ensemble methods (XGBoost, random forest, AdaBoost) and a multi-layer perceptron. Model-specific and model-agnostic explainable AI (XAI) techniques were applied to interpret model decisions, enhancing the transparency and clinical applicability of the proposed approach. The TabNet models demonstrated superior performance, achieving 94.06 % and 92.71 % accuracy in cross-validation for the complete and simplified models, respectively, and 91.00 % and 88.00 % accuracy on test sets. XAI analysis identified serum creatinine, cystatin C, sex, and age as the most influential factors in CKD stage classification. The proposed TabNet models offer a robust approach for early CKD severity detection in diabetes patients, potentially improving clinical decision-making and patient outcomes.
    Keywords:  CKD stage prediction; Chronic kidney disease (CKD); Deep learning; Diabetes; Machine learning; Prediction model
    DOI:  https://doi.org/10.1016/j.artmed.2025.103153
  8. Narra J. 2025 Apr;5(1): e2116
      Macrovascular complications, including stroke, cardiovascular disease (CVD), and peripheral vascular disease (PVD), significantly contribute to morbidity and mortality in individuals with type 2 diabetes mellitus (T2DM). The aim of this study was to evaluate the performance of artificial intelligence (AI) models in predicting these complications, emphasizing applicability in diverse healthcare settings. Following PRISMA guidelines, a systematic search of six databases was conducted, yielding 46 eligible studies with 184 AI models. Predictive performance was assessed using the area under the receiver operating characteristic curve (AUROC). Subgroup analyses examined model performance by outcome type, predictor data (lab-only, non-lab, mixed), and algorithm type. Heterogeneity was evaluated using I 2 statistics, and sensitivity analyses addressed outliers and study biases. The pooled AUROC for all AI models was 0.753 (95%CI: 0.740-0.766; I 2 = 99-99%)· Models predicting PVD achieved the highest AUROC (0.794), followed by cerebrovascular diseases (0.770) and CVD (0.741). Gradient-boosting algorithms outperformed others (AUROC: 0.789). Models with lab-only predictors had superior performance (AUROC: 0.837) compared to mixed (0.759) and non-lab predictors (0.714). External validations reported reduced AUROC (0.725), underscoring limitations in generalizability. AI models show moderate predictive accuracy for T2DM macrovascular complications, with laboratory-based predictors being key to performance. However, the limited external validation and reliance on high-resource data restrict implementation in low-resource settings. Future efforts should focus on non-lab predictors, external validation, and context-appropriate AI solutions to enhance global applicability.
    Keywords:  Artificial intelligence; cardiovascular disease; diabetic nephropathy and vascular disease; stroke; type 2 diabetes mellitus
    DOI:  https://doi.org/10.52225/narra.v5i1.2116
  9. Int J Mol Sci. 2025 Apr 22. pii: 3935. [Epub ahead of print]26(9):
      Type 1 diabetes (T1D) is an autoimmune condition characterized by the destruction of insulin-producing pancreatic beta cells, leading to lifelong insulin dependence and significant complications. Early detection of T1D is essential to delay disease onset and improve outcomes. Recent advancements in artificial intelligence (AI) and machine learning (ML) have provided powerful tools for predicting and diagnosing T1D. This systematic review evaluates the current landscape of AI/ML-based approaches for early T1D detection. A comprehensive search across PubMed, EMBASE, Science Direct, and Scopus identified 1447 studies, of which 10 met the inclusion criteria for narrative synthesis after screening and full-text review. The studies utilized diverse ML models, including logistic regression, support vector machines, random forests, and artificial neural networks. The datasets encompassed clinical parameters, genetic risk markers, continuous glucose monitoring (CGM) data, and proteomic and metabolomic biomarkers. The included studies involved a total of 49,172 participants and employed case-control, retrospective cohort, and prospective cohort designs. Models integrating multimodal data achieved the highest predictive accuracy, with area under the curve (AUC) values reaching up to 0.993 in sex-specific models. CGM data and plasma biomarkers, such as CXCL10 and IL-1RA, also emerged as valuable tools for identifying at-risk individuals. While the results highlight the potential of AI/ML in revolutionizing T1D risk stratification and diagnosis, challenges remain. Data heterogeneity and limited model generalizability present barriers to widespread implementation. Future research should prioritize the development of universal frameworks and real-world validation to enhance the reliability and clinical integration of these tools. Ultimately, AI/ML technologies hold transformative potential for clinical practice by enabling earlier diagnosis, guiding targeted interventions, and improving long-term patient outcomes. These advancements could support clinicians in making more informed, timely decisions, thus reducing diagnostic delays and paving the way for personalized prevention strategies in both pediatric and adult populations.
    Keywords:  early detection; evidence synthesis; machine learning; predictive modeling; type 1 diabetes
    DOI:  https://doi.org/10.3390/ijms26093935
  10. Clin Exp Med. 2025 May 10. 25(1): 151
      Although machine learning is frequently used in medicine for predictive purposes, its accuracy in diabetes-related amputation (DRA) remains unclear. From establishing the database until December 2024, we conducted a comprehensive search of PubMed, Web of Science (WoS), Embase, Scopus, Cochrane Library, Wanfang, and the China National Knowledge Index (CNKI). The pooled sensitivity, specificity, positive likelihood ratio (PLR), negative likelihood ratio (NLR), diagnostic odds ratio (DOR), area under the curve (AUC), and Fagan plot analysis were used to assess the overall test performance of machine learning. Moreover, subgroup analysis and meta-regression were performed to search for possible sources of heterogeneity. Finally, sensitivity analysis and Deeks' funnel plot asymmetry test were used to evaluate the stability and publication bias, respectively. In the end, seven publications were included in this meta-analysis. The overall pooled diagnostic data were as follows: sensitivity, 0.72 (95% CI 0.69-0.75); specificity, 0.89 (95% CI 0.84-0.93); PLR, 3.62 (95% CI 3.36-3.89); NLR, 0.32 (95% CI 0.30-0.35); DOR, 13.55 (95% CI 11.72-15.67). The AUC was 0.81 (95% CI 0.77-0.84). The Fagan plot analysis showed that the positive post-test probability is 62% and the negative post-test probability is 7%. Subgroup analysis and meta-regression showed that both the level of bias and the year of publication were sources of heterogeneity in sensitivity and specificity. Sensitivity analysis confirmed the robustness of the results after excluding three outlier studies. The Deeks' funnel plot suggests that publication bias has no statistical significance (P > 0.05). In summary, our results suggest the moderate accuracy of machine learning in predicting DRA.
    Keywords:  AUC; Accuracy; Diabetes-related amputation (DRA); Machine learning; Prediction
    DOI:  https://doi.org/10.1007/s10238-025-01697-w
  11. Public Health. 2025 May 09. pii: S0033-3506(25)00190-8. [Epub ahead of print]244 105744
       OBJECTIVES: To systematically review published studies on risk prediction models for patients with recurrent diabetic foot ulcers.
    STUDY DESIGN: Systematic review.
    METHODS: China National Knowledge Infrastructure (CNKI), Chinese Biomedical Literature Database (CBM), Wanfang Database, China Science and Technology Journal Database (VIP), PubMed, Web of Science, the Cochrane Library and Embase were searched from inception to November 5, 2023. Data from selected studies were extracted, including author, country, participants, study design, data source, sample size, outcome definition, predictors, model development and performance. The Prediction Model Risk of Bias Assessment Tool (PROBAST) checklist was used to assess the risk of bias and applicability.
    RESULTS: A total of 677 studies were retrieved, and after a screening process, eight predictive models from eight studies were included in this review. The studies utilized logistic regression, COX regression, and machine learning methods to develop risk prediction models for diabetic foot ulcer recurrence. The rate of diabetic foot ulcer recurrence was 20 %-41 %. The most commonly used predictors were HbA1c and DM duration. the reported area under the curve (AUC) ranged from 0.690 to 0.937. All studies were found to be at high risk of bias, mainly due to problems with outcome measures and poor reporting of analytic domains. the studies were not found to be at high risk of bias, mainly due to problems with outcome measures and poor reporting of analytic domains.
    CONCLUSIONS: Although the performance of the diabetic foot ulcer recurrence prediction models included in the studies was decent, all of them were found to be at high risk of bias according to the PROBAST checklist. Future studies should focus on developing new models with larger samples, rigorous study designs, and multicenter external validation.
    Keywords:  Diabetic foot ulcer; Recurrence; Risk prediction model; Systematic review
    DOI:  https://doi.org/10.1016/j.puhe.2025.105744