bims-aukdir Biomed News
on Automated knowledge discovery in diabetes research
Issue of 2026–07–12
24 papers selected by
Mott Given
  1. Construction and validation of an interpretable machine learning model for predicting diabetes risk in COPD patients.
  2. A comparative and interpretable machine learning framework for reliable diabetes risk prediction.
  3. Assessment of Diabetes Risk in Patients With Hepatitis B: A Machine Learning Approach Integrating 11 Inflammatory and Immune Indicators.
  4. Generative AI models for type 2 diabetes mellitus risk prediction.
  5. An XGBoost-based model for detecting undiagnosed type 2 diabetes using routine physical and lifestyle data from a multi-center Chinese population.
  6. A Lesion-Aware Detection Framework for Red Lesions in Retinal Fundus Images.
  7. Performance of AI in Predicting the Progression of Gestational Diabetes to Type 2 Diabetes: Systematic Review and Meta-Analysis.
  8. A CNN-injected transformer network with lesion reconstruction for multi-view diabetic retinopathy grading.
  9. Machine learning for predicting emergency department visits in patients with type 2 diabetes: A real-world, multi-institutional study.
  10. Artificial intelligence for detection, prediction, and severity classification of diabetic peripheral neuropathy: a systematic review.
  11. AI-Based Models for Diabetic Foot Ulcer Assessment: Scoping Review.
  12. Dual-decoder multi-task network with graph attention mechanism for OCT retinal layer and fluid segmentation.
  13. Development of a risk stratification tool for rapidly progressive diabetic retinopathy in type 2 diabetes.
  14. Interpretable deep learning to predict one year glycemic control in type 1 diabetes using real world data.
  15. Development and Internal Validation of a Machine Learning-Based Classification Model for Identifying Cognitive Frailty in Older Inpatients with Type 2 Diabetes Mellitus.
  16. Quantifying the Predictive Power of Social Determinants of Health in Cardiovascular Disease and Type 2 Diabetes Progression Using XGBoost: Retrospective Cohort Study.
  17. Exploring the relationship between psychological interventions and glycemic stability in youth with type 2 diabetes mellitus using machine learning methods.
  18. Discovery of hub genes linking oxidative stress to type 2 diabetic sarcopenia using single-cell sequencing and machine learning.
  19. Effect of RAS inhibitors on retinal microvasculature in DKD using AI-based ultra-wide-field images.
  20. Where You Live Shapes Diabetes Risk: Multilevel Determinants of Type 2 Diabetes in the East South Central United States.
  21. The association between estimated glucose disposal rate and self-reported diabetic retinopathy: evidence from two independent cohorts and machine learning.
  22. Evaluation of large language model-generated information in diabetes health patient education: a scoping review.
  23. Digital Management of Early-Onset Type 2 Diabetes: Empowerment, Challenges, and Future Outlook.
  24. Artificial Intelligence-Assisted Screening for Patients With Diabetic Retinopathy and Age-Related Macular Degeneration in Family Medicine and Geriatric and Gerontology Care: Protocol for a Pragmatic Randomized Clinical Trial.
  1. Sci Rep. 2026 Jul 09.
    Construction and validation of an interpretable machine learning model for predicting diabetes risk in COPD patients.
    Lingpin Pang, Siyan Xu, Yingxin Wang, Yiming Liu, Tao Huang, Qian Xian, Wenjia Lin, Haowen Pang, Zhirui Chen, Bozhi Zhong, Hui Miao, Hui Chen, Xishi Sun, Jie Sun, Xiaoting Huang.
      To construct and validate a machine learning (ML) model for predicting diabetes risk in COPD patients, enabling early and personalized intervention. Using data from the MIMIC-IV database, 49 variables were screened by LASSO and logistic regression. Six ML algorithms were constructed and internally validated on a 70%/30% split dataset. Model performance was assessed using multiple metrics, followed by external validation. Interpretability was achieved through SHAP analysis. All six ML algorithms demonstrated strong performance across training, testing, and validation sets. The LightGBM model achieved the best overall performance (AUC = 0.87). Feature importance analysis identified glucose, chronic kidney disease, and hyperlipidemia as the top three most important features for diabetes development in COPD patients. An interpretable ML-based risk prediction model for diabetes in COPD patients was constructed and validated. The LightGBM-based tool shows potential for supporting early personalized care and improving prognosis.
    Keywords:  Chronic obstructive pulmonary disease; Diabetes; Machine learning; Prediction model
    DOI:  https://doi.org/10.1038/s41598-026-61810-1
  2. Sci Rep. 2026 Jul 07.
    A comparative and interpretable machine learning framework for reliable diabetes risk prediction.
    Talha Farooq Khan, Mariyam Saeed, Majid Hussain, Lal Khan, Mohammad Zubair Khan, Adnan Nadeem, Turki Alghamdi.
      Diabetes mellitus is a common type of metabolic illness that is very common worldwide, and in most cases, it results in serious effects like heart disease, kidney disease, and blindness. Proper and early diagnosis of diabetes is essential to intervene on time and have better patient outcomes. Machine learning (ML) paradigms provide effective predictive modeling solutions to healthcare, but most of the current literature is limited due to imbalanced datasets, using a single training test split, and limited model interpretability, which diminish their clinical usability. This research paper has introduced a powerful and explainable ML model to predict diabetes based on the Pima Indians Diabetes Dataset acquired via Kaggle, which contains 768 patients with eight clinical variables and a binary response. To counter the class imbalance, the Synthetic Minority Over-sampling Technique (SMOTE) is used to create natural synthetic samples of the minority diabetic group that facilitate balanced learning without degrading the correlations between the features. Four classifiers, including Logistic Regression, Naive Bayes, AdaBoost, and XG Boost, are trained and tested. The stratified 10-fold cross-validation is used to provide a stable and generalizable model performance, as opposed to using only one data split. The measurement criteria are accuracy, precision, recall, and F1-score, especially for the minority diabetic class. The interpretation of the model is improved by the use of logistic regression coefficients and SHAP (SHapley Additive exPlanations) values, as they allow transparent identification of clinical features that are critical to making predictions. The results of the experiment show that the suggested framework attains an overall accuracy of approximately 94% on an unseen test set, with strong precision and recall of the minority class, thus proving that the combination of class balancing, cross-validation, and explainable ML results in the outcomes of reliable and clinically credible predictions. All performance results are evaluated on an untouched original test set, while SMOTE is applied strictly within cross-validation folds to prevent data leakage. Unlike many existing studies, the proposed framework ensures leakage-free validation, robust cross-validation, and integrated interpretability for clinically meaningful prediction. Although synthetic sampling improves minority class learning, the model is evaluated carefully to ensure generalization on real-world data. This paper indicates that a rigorously conducted methodology and interpretability in machine learning development are crucial in creating machine learning solutions in healthcare decision support, which is the pathway to real applications in diabetes risk assessment.
    Keywords:  Class imbalance; Diabetes prediction; Explainable AI (SHAP); Machine learning; Medical data analysis; Model interpretability; Stratified cross-validation
    DOI:  https://doi.org/10.1038/s41598-026-61038-z
  3. Diabetes Metab Res Rev. 2026 Jul;42(5): e70193
    Assessment of Diabetes Risk in Patients With Hepatitis B: A Machine Learning Approach Integrating 11 Inflammatory and Immune Indicators.
    Zhisheng Cai, Jingtao Liu, Wenjie Zhang, Yutong Lu, Yumeng Su, Tengyue Zhang, Yangyi Zhang, Xiaochi Wang, Linyi Ye, Ying Tao, Junjie Liu.
       BACKGROUND: Chronic hepatitis B virus (HBV) infection is associated with an increased risk of diabetes; however, early detection of diabetes remains challenging due to silent progression in early stages. Given that 30%-50% of diabetic patients develop severe complications such as cardiovascular disease, renal failure, and neuropathy, timely risk assessment is critical for prevention.
    METHODS: We developed a machine learning model using data from 14,287 HBV-positive adults across eight NHANES cycles (2003-2018) to identify diabetes. Eleven inflammation-immune composite indicators were integrated with demographic and biochemical parameters. Following LASSO variable selection and a 7:3 train-test split, seven algorithms were compared. Model performance was evaluated using the area under the ROC curve (AUC), calibration curve, and decision curve analysis (DCA).
    RESULTS: The Artificial Neural Network (ANN) emerged as the optimal model for diabetes risk assessment. It achieved an AUC of 0.83 (95% CI: 0.80-0.85) and an accuracy of 82% (95% CI: 0.79-0.85). SHAP analysis identified age and UHR as the most influential predictors, while HRR showed significant protective effects.
    CONCLUSION: This study presents a robust, interpretable model for diabetes risk assessment in HBV patients that integrates novel inflammation-immune composite indicators. These findings highlight the model's utility as a potential clinical tool for screening individuals with undiagnosed diabetes.
    Keywords:  SHAP; diabetes; hepatitis B; inflammatory immune factors; machine learning
    DOI:  https://doi.org/10.1002/dmrr.70193
  4. Health Syst (Basingstoke). 2026 ;15(1): 66-89
    Generative AI models for type 2 diabetes mellitus risk prediction.
    Elsa Sharu Johnson, Uma Gandhi, U Srinivasulu Reddy, Umapathy Mangalanathan.
      Diabetes, precisely Type II Diabetes Mellitus (T2DM), is a prevalent global chronic condition. This study focuses on improving the accuracy of predicting T2DM onset and risk by utilizing Generative Artificial Intelligence (GenAI) based synthetic data generation and innovative feature selection techniques. GenAI models such as Deep Tabular Augmentation (DTA) and Large Language Models (LLM) were utilized to address class imbalance and data scarcity of diabetes class for prediction. The Representative Instances-based Fuzzy Rough Set Feature Selection (FRS-RI) method was employed for optimal feature selection. Three diabetes datasets - Sylhet, Obesity, and Diagnostic Features - were employed. After FRS-RI feature selection and synthetic data generation, Machine Learning (ML), Ensemble Learning (EL), and Deep Learning (DL) models were trained on these datasets. The ML, EL, and DL models achieved impressive accuracy, precision, and recall scores: 95.19%, 0.96, and 0.94 for the Sylhet Dataset; 100%, 1.00, and 1.00 for the Obesity dataset; and 97.44%, 0.97, and 0.94 for the NIDDK-DF Dataset. The model's ability to generalize to new diabetic data was demonstrated by enhanced test accuracies of 98.37% and 97.33% obtained when the suggested techniques were applied to benchmark datasets such as PIMA and LMCH, respectively. Emphasis was also placed on model explainability to justify predictions for clinical presentation.
    Keywords:  Type II diabetes mellitus; feature selection; generative AI; large language models; machine learning and deep learning; onset and risk prediction
    DOI:  https://doi.org/10.1080/20476965.2025.2546904
  5. Front Med (Lausanne). 2026 ;13 1798509
    An XGBoost-based model for detecting undiagnosed type 2 diabetes using routine physical and lifestyle data from a multi-center Chinese population.
    Hui Xiao, Qian Xi, Ping Zeng, Jinjuan Hao, Qinghua He, Xiaoxia Wang, Chi Zhang.
       Introduction: This study aimed to develop and validate an interpretable machine learning model to identify individuals with undiagnosed type 2 diabetes (T2D) using data readily available from routine health checkups.
    Methods: In this retrospective study, we analyzed data from 12 tertiary hospitals in China. Following the application of inclusion and exclusion criteria, data from 11,382 individuals formed the training set for developing an XGBoost model, which was optimized using 5-fold cross-validation. An independent test set of 1,026 individuals from the same multi-center data source was used for internal validation. Model performance was primarily assessed using the area under the receiver operating characteristic curve (AUC).
    Results: The final model incorporated 12 predictors. Fasting blood glucose was the most influential predictor (50.6%), followed by creatinine (6.6%), triglyceride (5.6%), age (5.1%), and low-density lipoprotein (5.0%). On the independent test set, the model achieved an AUC of 77.2% (95%CI: 70.3%-84.1%).
    Conclusion: The XGBoost model demonstrated moderate predictive performance for T2D risk using routine checkup data. This approach shows potential for integration into clinical practice as an assistive screening tool, enabling automated risk profiling during standard health examinations. By flagging high-risk individuals, it can support clinicians in decision-making regarding further diagnostic testing. Future work should focus on external validation and prospective implementation studies.
    Keywords:  XGBoost; health checkup; machine learning; risk prediction; type 2 diabetes
    DOI:  https://doi.org/10.3389/fmed.2026.1798509
  6. J Imaging Inform Med. 2026 Jul 07.
    A Lesion-Aware Detection Framework for Red Lesions in Retinal Fundus Images.
    Pritam Mandal, Swagata Kundu, Aatreya Sengupta, Raka Goswami, Sushmita Mitra, Ramachandran Rajalakshmi, Thyparambil Aravindakshan PramodKumar, Sambuddha Ghosh, Ashis Kumar Dhara.
      Diabetic retinopathy (DR) is a serious eye condition caused by damage to the retina due to prolonged diabetes. Among the first signs of DR are microaneurysms and hemorrhages, collectively called red lesions. Detecting these lesions is crucial in preventing vision loss. This paper presents a refined feature attention (RFA) module that refines features from the feature pyramid network (FPN) layers before they are passed to the detection head. Dataset-specific adaptive anchor ratios are derived using a quantile-based multi-stage clustering strategy to better align the anchors with lesion geometry. A lesion-aware sampling strategy based on the morphological characteristics of red lesions is proposed to balance the batch during training for datasets with a wide range of spatial distributions and structural complexities. The network's performance was evaluated across three standard retinal image datasets: Messidor-A, SaNMoD, and E-Ophtha MA. On the Messidor-A and SaNMoD datasets, the proposed method achieved FROC scores of 0.4650 and 0.4429, respectively. Furthermore, evaluation on the E-Ophtha MA dataset yielded an FROC score of 0.549, demonstrating the robustness and generalization capability of the proposed model. The proposed network achieved a precision of 73.6%, a recall of 69.0%, and an F1-score of 71.2% on the Messidor-A dataset, with false positives per image (FPI) of 2.802 and an average inference time of 45.24 ms per image. The model requires 128 GFLOPs and has 36.35 million parameters. Experiments show that combining the attention module with adaptive anchoring and lesion-aware sampling improves the model's ability to detect red lesions.
    Keywords:  Clustering; Diabetic retinopathy; Feature pyramid; Lesion-aware sampling; Red lesions
    DOI:  https://doi.org/10.1007/s10278-026-02096-7
  7. J Med Internet Res. 2026 Jul 09. 28 e87882
    Performance of AI in Predicting the Progression of Gestational Diabetes to Type 2 Diabetes: Systematic Review and Meta-Analysis.
    Alaa Abd-Alrazaq, Shahira Padinharepattel Mohamed, Mohannad Alajlani, Aliya Tabassum, José Manuel Ordóñez-Mena, Shehel Yoosuf, Mais Alkhateeb, Arfan Ahmed, Mohammed Bashir, Junaid Qadir, Ali AlSanousi, Javaid Sheikh.
       Background: Gestational diabetes mellitus (GDM) significantly increases the risk of developing type 2 diabetes mellitus (T2DM) post partum, with up to half of affected women progressing within a decade. Early identification of high-risk individuals is critical for implementing preventive interventions. Artificial intelligence (AI) offers enhanced predictive capabilities that can substantially enhance the prevention of postpartum diabetes.
    Objective: This systematic review and meta-analysis aimed to evaluate the performance of AI models in predicting the progression from GDM to T2DM or prediabetes.
    Methods: A total of 7 databases (MEDLINE, Embase, Scopus, Web of Science, IEEE Xplore, ACM Digital Library, and Google Scholar) were systematically searched from inception through September 12, 2025, supplemented by backward and forward reference screening and biweekly alerts to capture newly published studies. This review included peer-reviewed English-language studies that applied AI algorithms to predict T2DM or prediabetes among women with previous GDM. Eligible studies focused on human participants; reported performance metrics (eg, accuracy, sensitivity, and specificity); and excluded non-AI models, animal studies, reviews, protocols, abstracts, and non-English publications. Moreover, 2 reviewers independently conducted study selection, data extraction, and risk of bias assessment using the PROBAST (Prediction Model Risk of Bias Assessment Tool)+AI tool. Pooled estimates were computed using random-effects meta-analysis models.
    Results: In total, 10 studies met the inclusion criteria, of which 8 were eligible for meta-analysis. The reviewed studies spanned from 2011 to 2025 and were conducted across 7 countries, predominantly in the United States (3/10, 30%). Most publications were journal articles (9/10, 90%), and retrospective designs (6/10, 60%) were slightly more common than prospective designs (4/10, 40%). AI models demonstrated high predictive performance for T2DM, with pooled accuracy of 0.85 (95% CI 0.79-0.90; prediction interval [PI] 0.64-0.98), sensitivity of 0.89 (95% CI 0.81-0.95; PI 0.63-1.00), specificity of 0.88 (95% CI 0.81-0.93; PI 0.67-0.99), F1-score of 0.80 (95% CI 0.75-0.85; PI 0.68-0.93), and area under the curve of 0.86 (95% CI 0.77-0.91; PI 0.54-0.97). However, AI performance for prediabetes prediction was modest (area under the curve=0.69, 95% CI 0.60-0.77). Subgroup analyses showed that random forest, decision tree, logistic regression, and naïve Bayes models performed comparably. Fasting plasma glucose and BMI were the most identified significant predictors in the included studies.
    Conclusions: AI models show potential in predicting T2DM after GDM. However, evidence remains limited by small sample sizes, high heterogeneity, lack of external validation, and high risk of bias. Our findings have important implications for digital health, supporting the integration of AI-driven risk prediction into electronic health record systems and postpartum care pathways to enable early identification, targeted prevention, and improved long-term outcomes. Future research should use large, diverse cohorts, integrate multidimensional data, adopt standardized reporting frameworks, and encourage open-access data sharing.
    Keywords:  artificial intelligence; diabetes mellitus; gestational diabetes; machine learning; meta-analysis; prediabetes; systematic review
    DOI:  https://doi.org/10.2196/87882
  8. Med Image Anal. 2026 Jul 04. pii: S1361-8415(26)00247-1. [Epub ahead of print]113 104178
    A CNN-injected transformer network with lesion reconstruction for multi-view diabetic retinopathy grading.
    Yongting Hu, Haoran Li, Qihao Xu, Yuxin Lin, Jiahua Shi, Jun Shen, Yong Xu, Xiaoyan Dou.
      The type and quantity of lesions are critical determinants in the assessment of diabetic retinopathy (DR) grading. Since multi-view fundus images provide a broader field of view and capture more lesions, multi-view DR grading has garnered increasing attention in recent years. However, existing multi-view methods either only focus on fundus feature extraction, or only take the lesion map as a part of the input, failing to fully leverage the comprehensive lesion information. Moreover, the significant variation in lesion size and their scattered distribution present substantial challenges for effective information learning. To address these issues, this paper proposes a CNN-injected transformer network with Lesion Reconstruction for Multi-View DR grading (LRMVDR), which utilizes lesion maps twice to fully exploit lesion information. Specifically, to tackle the large-scale variations and widespread distribution of lesions, the lesion maps are concatenated with the fundus images and then input into the local and global branches for extracting hierarchical global-local features. Adapters are designed to inject CNN features into the Transformer between the two branches, significantly enhancing the integration of multi-scale global and local features. Additionally, a dedicated lesion reconstruction branch is employed to explicitly extract lesion features. These features are subsequently fused with those from the local branch via a wavelet enhancement module, enabling Interactive fusion of frequency domain information and spatial domain information. Extensive experiments on large public datasets demonstrate the effectiveness and competitiveness of the proposed method. Our code is available at https://github.com/HuYongting/LRMVDR.
    Keywords:  Diabetic retinopathy; Feature injection; Lesion reconstruction; Multi-view learning
    DOI:  https://doi.org/10.1016/j.media.2026.104178
  9. PLoS One. 2026 ;21(7): e0352342
    Machine learning for predicting emergency department visits in patients with type 2 diabetes: A real-world, multi-institutional study.
    Sunyoung Kim, Hyunji Sang, Jaeyu Park, Selin Woo, Eun-Hee Cho, Chong Hwa Kim, Dae Jung Kim, Chang-Won Jeong, Tae Sun Park, You-Cheol Hwang, Hyunjung Lim, Zio Kim, Hyejin Kang, Dong Keon Yon, Sang Youl Rhee.
       BACKGROUND: Patients with type 2 diabetes mellitus (T2DM) prone to acute diabetic complications are at high risk for emergency department (ED) visits, which often precede hospitalization and mortality. Identifying these high-risk phenotypes before deterioration is critical for preventative care. We developed machine learning (ML) models using large-scale, real-world electronic medical records, including prescription data, to predict the possibility of ED visits in patients with T2DM and support proactive interventions in primary care settings.
    METHODS: We analyzed the electronic health record data of five independent institutions, creating a comprehensive dataset of 220,720 patients. The data included dynamic clinical parameters such as vital signs, laboratory results, and prescription histories. The cohort was randomly split into a training set (n = 176,576) and a test set (n = 44,144). The primary outcome was the first ED visit. We developed multiple ML models using an automated ML framework and optimized them using hyperparameter tuning of the training set. Model performances were evaluated using the area under the receiver operating characteristic (AUROC) curve, and feature importance was analyzed using SHAP values to ensure interpretability.
    RESULTS: Among the screened population, 49,770 (22.6%) experienced at least one ED visit, distributed proportionally across the training and test datasets. The CatBoost model demonstrated superior predictive performance, achieving an AUROC of 0.87 (95% CI, 0.862-0.871) on the test dataset. The model identified modifiable risk factors as key predictors; Diastolic blood pressure was the most significant variable, followed by serum creatinine and systolic blood pressure.
    CONCLUSIONS: This ML-based predictive model can accurately identify high-risk patients with T2DM who are likely to visit the ED based on readily available clinical variables. By enabling healthcare providers to shift from reactive treatment to proactive risk management, it has the potential to reduce the burden of ED visits due to acute complications in T2DM.
    DOI:  https://doi.org/10.1371/journal.pone.0352342
  10. BMC Med Inform Decis Mak. 2026 Jul 10.
    Artificial intelligence for detection, prediction, and severity classification of diabetic peripheral neuropathy: a systematic review.
    Tay Jia Ying, Ang Yee Gary.
       BACKGROUND: Diabetic peripheral neuropathy (DPN) is a common complication of diabetes and an important contributor to foot ulceration and lower limb amputation. Early detection remains challenging because conventional screening methods are often subjective, resource-intensive, and insensitive to subclinical disease. Artificial intelligence (AI) has increasingly been applied to support DPN assessment, but its applications across different clinical tasks have not been clearly synthesised. This systematic review evaluates AI applications for DPN detection, prognostic prediction, risk stratification, and severity classification.
    METHODS: A systematic literature search was conducted across PubMed, Scopus, and IEEE Xplore for studies published from January 2021 to 1 September 2025. Studies developing or validating AI models for DPN assessment using patient datasets were included. Study selection was performed by one reviewer, with a second reviewer contributing to uncertain inclusion decisions. Data extraction and risk-of-bias assessment were performed by one reviewer and verified by another using the Prediction Model Risk of Bias Assessment Tool (PROBAST). Due to heterogeneity in clinical tasks, data types, outcome definitions, and validation approaches, findings were synthesised narratively.
    RESULTS: Twenty-six studies were included and stratified by clinical task: diagnostic detection of existing DPN, prognostic prediction or risk stratification, and severity classification. Diagnostic detection studies mainly used corneal confocal microscopy, plantar pressure or gait-based assessments, nerve conduction studies, electromyography, and selected clinical variables. Prognostic prediction and risk stratification studies relied mainly on electronic medical record variables, including age, diabetes duration, glycated haemoglobin, renal markers, inflammatory markers, and cardiovascular risk factors. Severity classification studies used electrophysiological measures, structured clinical examination findings, and symptom-based scores. Across tasks, age, diabetes duration, and glycaemic control were recurrent predictors, while imaging, gait, plantar pressure, and electrophysiological features were more task-specific. Reported internal performance was often high, but most studies used small or single-centre datasets, heterogeneous reference standards, and limited reporting of predictor importance. External validation was uncommon, and risk of bias was frequently high, particularly in the analysis domain.
    CONCLUSIONS: AI approaches for DPN assessment show promise, but interpretation depends on task. Future studies should develop task-specific models, report predictor importance transparently, and undertake external and prospective validation before routine clinical implementation.
    Keywords:  Artificial intelligence; Diabetic peripheral neuropathy; Machine learning; Prognostic prediction; Risk stratification; Severity classification
    DOI:  https://doi.org/10.1186/s12911-026-03676-x
  11. JMIR Diabetes. 2026 Jul 08. 11 e77925
    AI-Based Models for Diabetic Foot Ulcer Assessment: Scoping Review.
    Muhamad Zulfiqar, Saldy Yusuf, Muhammad Jufri Taming, Herlina Burhan.
       Background: Diabetic foot ulcers (DFU) are serious complications of diabetes that contribute substantially to morbidity, mortality, and health care burden. Accurate and timely wound assessment is essential for effective DFU management; however, conventional assessment methods are limited by subjectivity, time constraints, and interobserver variability.
    Objective: This scoping review aimed to map and synthesize evidence regarding the development and application of artificial intelligence (AI)-based models for DFU assessment.
    Methods: A scoping review was conducted following the Arksey and O'Malley framework and reported according to the PRISMA-ScR (Preferred Reporting Items for Systematic Reviews and Meta-Analyses Extension for Scoping Reviews) guidelines. Literature searches were performed in PubMed, ProQuest, and Scopus for studies published between 2014 and 2026. Study selection and data charting were conducted independently by two reviewers using predefined inclusion criteria based on the PCC (population, concept, context) framework. Extracted data were synthesized narratively and categorized according to major AI application domains.
    Results: A total of 654 records were identified, of which 46 studies met the inclusion criteria. The included studies predominantly focused on image segmentation, diagnostic classification, and risk prediction or monitoring of DFUs. Convolutional neural networks were the most commonly applied models, with performance evaluated using metrics such as accuracy, Dice similarity coefficient, and area under the curve. Most studies relied on retrospective, single-center datasets, with limited external validation and minimal real-world clinical implementation.
    Conclusions: AI-based models demonstrate strong potential to enhance DFU assessment and monitoring by improving accuracy and efficiency. However, significant gaps remain in terms of dataset diversity, external validation, and integration into clinical workflows. Future research should prioritize prospective validation, standardized datasets, and real-world implementation to support safe and effective clinical adoption.
    Keywords:  artificial intelligence; deep learning; diabetic foot ulcers; diabetic wound; scoping review
    DOI:  https://doi.org/10.2196/77925
  12. BMC Ophthalmol. 2026 Jul 04.
    Dual-decoder multi-task network with graph attention mechanism for OCT retinal layer and fluid segmentation.
    Xuan Dong, Idowu Paul Okuwobi.
       BACKGROUND: Diabetic macular edema (DME) is a leading cause of visual impairment and blindness among the diabetic population, and leads to abnormal retinal morphology, distorted layer boundaries and blurred structures in optical coherence tomography (OCT) images. Accurate segmentation of retinal layers and pathological fluid regions is critical for clinical diagnosis, but remains challenging due to irregular fluid distribution and low boundary contrast. This study aims to develop an effective segmentation method to jointly extract retinal layers and fluid regions for assisting clinical screening.
    METHODS: A novel dual-decoder multi-task network with graph attention mechanism was proposed for joint segmentation. A primary decoder completed region segmentation, while an auxiliary decoder focused on boundary detection. A cross-decoder spatial attention module was designed for bidirectional feature interaction, and a global reasoning module was embedded to capture long-range anatomical dependencies. Experiments were conducted on the public Duke DME dataset with five-fold subject-independent cross-validation, and paired t-tests were adopted for statistical significance analysis.
    RESULTS: The proposed method outperformed comparative mainstream segmentation models in overall and category-wise evaluation. It achieved stable accuracy in normal retinal layer segmentation and obtained competitive performance in identifying fluid regions, effectively reducing the interference of pathological changes and improving boundary consistency of segmentation results.
    CONCLUSIONS: The proposed method enables accurate joint segmentation of retinal layers and fluid regions. It provides a reliable automated analysis tool for diabetic macular edema, and can serve as an effective auxiliary reference for routine clinical screening and quantitative evaluation.
    Keywords:  Cross-decoder spatial attention; Diabetic macular edema; Optical coherence tomography; Retinal fluid segmentation; Retinal layer segmentation
    DOI:  https://doi.org/10.1186/s12886-026-05081-4
  13. Front Endocrinol (Lausanne). 2026 ;17 1837957
    Development of a risk stratification tool for rapidly progressive diabetic retinopathy in type 2 diabetes.
    Aiping Gu, Zhichao Yan, Yi Wu, Yanying Li, Renlong Liang, Xiaodi Tang, Mengke Li.
       Background: The progression of rapidly progressive diabetic retinopathy (PDR) in type 2 diabetes mellitus (T2DM) is characterized by substantial inter-individual variability. To develop and validate a nomogram for individualized risk prediction and stratification of rapidly progressive PDR in T2DM by incorporating diabetes duration, glycated hemoglobin (HbA1c), 24-hour urinary protein quantification, growth differentiation factor 15 (GDF15), Diabetic Retinopathy Severity Scale (DRSS) grade, and foveal avascular zone area.
    Methods: This retrospective study enrolled 342 patients with T2DM (1999 WHO criteria), randomly assigned to training (n=240) and validation (n=102) sets (7:3 ratio). Baseline demographic, clinical, metabolic, renal, inflammatory biomarker, and ophthalmic imaging data were collected. Predictive variables were selected via univariate analysis and least absolute shrinkage and selection operator (LASSO) regression. Independent predictors identified by multivariable logistic regression were incorporated into a nomogram. For comparison, Random Forest, multivariable logistic regression, and Gradient Boosting Machine models were also developed. Model performance was assessed using the area under the receiver operating characteristic curve (AUC-ROC), calibration curves, and decision curve analysis (DCA).
    Results: Univariate analysis identified six significant factors (all P < 0.05): diabetes duration, HbA1c, 24-hour urinary protein quantification, GDF15, DRSS grade, and foveal avascular zone area. LASSO regression retained all six, and multivariable logistic regression confirmed them as independent risk factors for rapidly progressive DR in T2DM (all P < 0.05). Three machine learning models were constructed. The Random Forest model achieved the numerically highest validation AUC (0.780) compared with Gradient Boosting Machine (0.741) and multivariable logistic regression (0.698), though the DeLong test showed no statistically significant difference between Random Forest and Gradient Boosting Machine (P = 0.38). Calibration curves showed good consistency between predicted and observed probabilities. DCA indicated high clinical net benefit of the model at 0.1-0.8 threshold probability vs other models and extreme strategies.
    Conclusion: A novel risk prediction model for rapidly progressive PDR in T2DM was developed and validated by integrating multidimensional parameters. Demonstrating favorable discrimination, calibration, and clinical utility, this model provides a promising tool for early identification of high-risk individuals and optimization of personalized intervention strategies.
    Keywords:  machine learning; nomogram; prediction model; rapidly progressive diabetic retinopathy; risk stratification; type 2 diabetes mellitus
    DOI:  https://doi.org/10.3389/fendo.2026.1837957
  14. Sci Rep. 2026 Jul 04.
    Interpretable deep learning to predict one year glycemic control in type 1 diabetes using real world data.
    Jose Tapia-Galisteo, Francisco Javier Somolinos-Simón, M Elena Hernando, Gema García-Sáez.
      Type 1 diabetes (T1D) involves long-term health risks and challenges in individualizing therapeutic strategies. Meeting glycemic targets is a reliable indicator of effective diabetes management and positive prognosis. This study develops a clinically interpretable predictive model of 1-year glycemic control-defined as a binary outcome based on HbA1c values-using Real-World Data from 8999 T1D patients. A 1-year horizon is clinically meaningful, as annual reassessment aligns with standard care guidelines and supports timely treatment adjustments and complication screening. Deep Learning techniques are evaluated for discrimination and calibration. Various feature subsets, calibration methodologies, and sampling strategies for unbalanced outcomes are compared. The best-performing model includes 12 features, encompassing socio-demographics, clinical variables, associated complications, and pharmacological treatment. The scaling-binning calibration technique achieved the best calibration performance. The final model yielded an area under the receiver operating characteristic curve of 0.870, an F1-score of 0.789, and calibration errors between 0.014 and 0.038. Sampling techniques did not outperform unbalanced models followed by calibration. To enhance interpretability, a graphical representation quantifies the contribution of each variable to the patient's risk score. Combining strong predictive accuracy, calibration, and interpretability, the model may help clinicians make individualized decisions, intensify care for high-risk patients, and optimize healthcare resource allocation.
    Keywords:  Calibration; Deep learning; Diabetes; Discrimination; Glycemic control prediction; Interpretability
    DOI:  https://doi.org/10.1038/s41598-026-59937-2
  15. Clin Interv Aging. 2026 ;21 594752
    Development and Internal Validation of a Machine Learning-Based Classification Model for Identifying Cognitive Frailty in Older Inpatients with Type 2 Diabetes Mellitus.
    Sijing Wang, Tingting Tan, Qingqing Wang, Cong Xu, Chongyao Yang, Miaoqin Wang, Kaili Dai.
       Background: Cognitive frailty (CF) is prevalent in older adults with type 2 diabetes mellitus (T2DM) and significantly increases risks of adverse outcomes. Early detection is essential for potential reversibility, requiring efficient bedside identification tools.
    Methods: This study included 523 consecutive older adults (aged 65 years or older) diagnosed with T2DM at a tertiary care center in Sichuan, China. Within a health ecological framework, 35 candidate variables were collected. Multiple imputation was used for missing data. LASSO regression was applied to the full candidate set to select key features. Six algorithms (logistic regression, decision tree, random forest, support vector machine, XGBoost, and stacking) were trained to identify CF with a 70/30 split (training/test). Performance was evaluated on the test set using AUC, AP, accuracy, sensitivity, specificity, and calibration; SHAP was used to assess model interpretability.
    Results: Mean age was 74.3±6.4 years, and CF prevalence was 41.5%. LASSO identified eight indicators: age, cognitive activities, physical exercise, diabetes complications, nutritional status, depressive symptoms, perceived social support, and insomnia symptoms. On the test set, logistic regression achieved the best performance for identification (AUC 0.947, AP 0.936, accuracy 0.879, sensitivity 0.838, specificity 0.910), with good calibration. SHAP analysis revealed that older age, lower frequency of physical exercise, higher scores for depressive symptoms, lower social support, poorer nutritional status, a greater number of diabetes complications, lack of cognitive activities, and higher scores for insomnia symptoms were the major contributors to a higher likelihood of CF.
    Conclusion: In this single-center hospital-based sample, a parsimonious model using eight readily obtainable features demonstrated strong internal validity for identifying CF in older inpatients with T2DM. This tool may facilitate early screening and clinical decision-making. Prospective studies are warranted to confirm its clinical utility.
    Keywords:  cognitive frailty; health ecological theory; identification; machine learning; older adults; type 2 diabetes mellitus
    DOI:  https://doi.org/10.2147/CIA.S594752
  16. JMIR Med Inform. 2026 Jul 09. 14 e80377
    Quantifying the Predictive Power of Social Determinants of Health in Cardiovascular Disease and Type 2 Diabetes Progression Using XGBoost: Retrospective Cohort Study.
    Hielke Muizelaar, Marcel Haas, Maarten van Aken, Rimke Vos, Marco Spruit.
       Background: Cardiovascular diseases (CVDs) and type 2 diabetes (DM2) are influenced not only by biomedical risk factors but also by social determinants of health (SDOH). While the inclusion of SDOH in predictive models is increasingly advocated, few studies have quantified their specific contribution in a high-risk clinical cohort using robust statistical and machine-learning approaches.
    Objective: This study aims to quantify the added predictive value of SDOH in predicting CVD or DM2 disease onset within 5 years, within 10 years, and at any time during follow-up among individuals already at elevated risk and to compare this added value across multiple modeling setups and frameworks.
    Methods: We used a large, linked dataset of over 58,000 inclusion events from the Extramural Leiden University Medical Center Academic Network data warehouse in the Netherlands, combining structured coded diagnosis and medication records from general practitioners with individual-level socioeconomic data from Statistics Netherlands. Individuals aged 30 years and older without prior DM2 or CVD were followed to assess disease progression. We trained Cox proportional hazards (CPH) and Extreme Gradient Boosting (XGBoost) models to predict progression to DM2 or CVD within 5 and 10 years and overall. All analyses were performed using the R programming language. Experiments included comparisons of Systematic Coronary Risk Evaluation 2, CPH, and XGBoost models; evaluation of time-bound and survival-based formulations; and quantification of SDOH impact using feature subset XGBoost models and Shapley additive explanations (SHAP)-based importance.
    Results: For the 5-year prediction of CVD or DM2, the combined XGBoost model using biomedical and SDOH predictors achieved an area under the receiver operating characteristic curve (AUC) of 0.738, significantly outperforming the biomedical-only model (AUC=0.728; P=.01) and the SDOH-only model (AUC=0.691; P<.001). For 10-year CVD prediction, XGBoost achieved an AUC of 0.729, outperforming CPH (AUC=0.718; P=.02) and Systematic Coronary Risk Evaluation 2 (AUC=0.697; P<.001). For overall event prediction, XGBoost again performed best (AUC=0.719), significantly higher than CPH (AUC=0.704; P<.001). SHAP analyses showed that biomedical predictors contributed most strongly on a per-feature basis, while a subset of SDOH variables, particularly income- and benefit-related indicators, provided complementary predictive signal and ranked among the most influential predictors.
    Conclusions: Incorporating SDOH improved the prediction of CVD and DM2 onset in a clinically defined high-risk cohort. Across hundreds of linked predictors, SDOH provided measurable incremental discrimination beyond biomedical risk factors, and income- and benefit-related variables ranked among the most influential features. SHAP analyses indicated that this added value was largely driven by a limited subset of highly informative social predictors. These findings support integrating structured SDOH into clinically actionable risk stratification models.
    Keywords:  XGBoost; biomedical risk factors; cardiometabolic disease; cardiovascular disease; machine learning; quantification; risk prediction; social determinants of health; type 2 diabetes
    DOI:  https://doi.org/10.2196/80377
  17. Ann Pediatr Endocrinol Metab. 2026 Jun;31(3): 203-205
    Exploring the relationship between psychological interventions and glycemic stability in youth with type 2 diabetes mellitus using machine learning methods.
    Jisun Park, Young Se Kwon, Ji-Eun Lee, Dong Hyun Kim.
      
    DOI:  https://doi.org/10.6065/apem.2550248.124
  18. PLoS One. 2026 ;21(7): e0352753
    Discovery of hub genes linking oxidative stress to type 2 diabetic sarcopenia using single-cell sequencing and machine learning.
    Guangwen Zhu, Kai Zou, Yi Liang, Liting Xie, Qiu Chen.
      Type 2 diabetes mellitus (T2DM) and sarcopenia demonstrate a significant comorbidity, particularly in the elderly, yet the molecular mechanisms linking them, especially through oxidative stress, remain incompletely understood. This study aimed to identify oxidative stress-related hub genes involved in T2DM-associated sarcopenia (T2DS) by integrating single-cell RNA sequencing (scRNA-seq) and bulk RNA-seq data with machine learning. We analyzed scRNA-seq datasets (GSE244515, GSE268953) to characterize cellular heterogeneity and bulk RNA-seq datasets (GSE202295, GSE226151) for differential expression. Cell type annotation revealed key involvement of neuromuscular junctions and myofibers. Functional enrichment analyses highlighted pathways like the proteasome, TNF signaling, and ubiquitin-mediated proteolysis. From an initial set of oxidative stress-related genes, a comprehensive machine learning framework comprising 127 algorithm combinations was employed. The Lasso+Stepglm[both] model identified 12 candidate genes. Subsequent Protein-Protein Interaction (PPI) network analysis refined this to seven core hub genes: TNFRSF1B, PSMA2, UBE2D1, UBE2N, HSP90AA1, RAD23A, and DNAJB1. These genes are functionally interconnected, primarily implicating TNFRSF1B-mediated inflammatory signaling that activates the ubiquitin-proteasome system, leading to enhanced protein degradation-a key pathway in muscle atrophy. ROC curve analysis confirmed the strong diagnostic value of these hub genes across training, test, and external validation sets. Our findings systematically reveal novel oxidative stress-related hub genes and mechanisms in T2DS, providing potential biomarkers and therapeutic targets for this debilitating condition.
    DOI:  https://doi.org/10.1371/journal.pone.0352753
  19. Front Endocrinol (Lausanne). 2026 ;17 1840180
    Effect of RAS inhibitors on retinal microvasculature in DKD using AI-based ultra-wide-field images.
    Yan Liu, Xinyu Zhao, Haiyun Wang, Jie Ma, Peng Xia, Xingwang Gu, Youxin Chen, Limeng Chen.
       Background: Diabetic kidney disease (DKD) and diabetic retinopathy (DR) represent two major microvascular complications of diabetes mellitus (DM). Previous studies have suggested that renin-angiotensin system inhibitors (RASi) exert protective effects on both DKD and DR. However, their specific impact on retinal microvascular parameters (RMPs), as well as the association between changes in fundus microvasculature and alterations in renal clinical parameters, remains unclear. This pilot study aimed to quantitatively assess the short-term effects of RASi on retinal microvasculature in patients with DKD using an artificial intelligence (AI)-based analysis of ultra-wide-field (UWF) fundus images.
    Methods: In this prospective cohort study, 27 patients with DKD were enrolled between July 2023 and September 2024. UWF fundus images were acquired at baseline and 12 weeks after initiation of RASi therapy. A validated deep learning AI model was employed to segment retinal vessels and quantify RMPs, including fractal dimension (Df) and tortuosity (TORT), in both the central and peripheral retinal regions. Statistical analyses for pre- and post-treatment comparisons were performed using a linear mixed-effects model with patient ID as a random intercept. or a Wilcoxon signed-rank test, as appropriate. Changes in these parameters post-treatment were analyzed and correlated with alterations in clinical renal indicators.
    Results: Among the enrolled patients, 21 patients (77.8%) were male, with a mean age of 55.7 ± 14.2 years, a mean diabetes duration of 9.9 ± 6.6 years, a baseline estimated glomerular filtration rate (eGFR) of 73.7 ± 18.5 mL/min/1.73 m², and a median proteinuria of 0.57 (0.25, 0.97) g/24h. Fifteen patients (55.6%) had diabetic retinopathy (DR). After 12 weeks of RASi treatment, significant decreases were observed within the UWF images in venous Df (adjusted p = 0.0389) for the overall cohort, and venous TORT (adjusted p = 0.0496) for No-DR and NPDR groups. These significant changes were not observed in parameters derived from the central retinal region.
    Conclusion: RASi therapy might be associated with retinal peripheral venous alterations, providing clues to a vascular- and topography-specific therapeutic response and shedding light on a potential imaging biomarker for diabetes management.
    Keywords:  artificial intelligence model; diabetic kidney disease; diabetic retinopathy; retinal microvascular parameters; ultra-wide-field fundus images
    DOI:  https://doi.org/10.3389/fendo.2026.1840180
  20. J Racial Ethn Health Disparities. 2026 Jul 09.
    Where You Live Shapes Diabetes Risk: Multilevel Determinants of Type 2 Diabetes in the East South Central United States.
    Ferial Ahmadi, Maryam Karimi, Rouzbeh Nazari, Alireza Eskandarinejad, Mohammad Reza Nikoo.
      Type 2 diabetes melittus (T2DM) remains a major public health challenge in the United States, particularly in the east south central regions. his study investigated behavioral, socioeconomic, and built environment determinants of county-level diabetes prevalence across Alabama, Kentucky, Mississippi, and Tennessee using linear mixed-effects and Random Forest models. The results indicated that food insecurity (β = 0.057, p < 0.001), smoking (β = 0.151, p < 0.001), poverty (β = 0.027, p < 0.001), the percentage of uninsured individuals (β = 0.072, p < 0.001), and the percentage of Black residents (β = 0.058, p < 0.001) were significantly associated with higher T2DM prevalence. Significant interactions were observed between obesity and physical inactivity (β = 0.0030, p < 0.001), obesity and alcohol consumption (β = - 0.0120, p < 0.001), and obesity and social deprivation (β = - 0.00081, p < 0.001), indicating that the association between obesity and diabetes prevalence varied according to behavioral and socioeconomic conditions. The Random Forest model demonstrated strong predictive performance (R2 = 0.884) and identified binge drinking, the percentage of Black residents, physical inactivity, poverty, obesity, food insecurity, and social deprivation as the most influential predictors. The consistency in findings across both multilevel and machine learning approaches highlights the combined influence of behavioral, socioeconomic, and environmental factors in shaping regional diabetes disparities and underscores the need for targeted public health interventions to reduce diabetes-related health inequities in the U.S. east south regions.
    Keywords:  Built environment; Machine learning; Spatial analysis; Type 2 diabetes; Urban health
    DOI:  https://doi.org/10.1007/s40615-026-03110-y
  21. Front Endocrinol (Lausanne). 2026 ;17 1853329
    The association between estimated glucose disposal rate and self-reported diabetic retinopathy: evidence from two independent cohorts and machine learning.
    Chaofeng Yuan, Yue Hao, Jianghui Wang, Chuanxi Wang, Zhengxuan Jiang.
       Background: Insulin resistance plays a key role in the pathogenesis of diabetic retinopathy (DR). Although established insulin resistance markers have been shown to predict a variety of complications, the association between the estimated glucose disposal rate (eGDR) and prevalence of DR remains incompletely characterized. This study aims to examine the relationship between eGDR and DR prevalence.
    Methods: This cross-sectional study analyzed complete participant data (N = 1, 536) from the 2007-2018 National Health and Nutrition Examination Survey (NHANES) for all relevant information. The relationship between the insulin resistance index and self-reported DR prevalence was evaluated by using multivariate logistic regression and a restricted cubic spline (RCS) model. Subgroup analysis was conducted to assess heterogeneity across groups, and two sensitivity analyses were performed to assess the robustness of the results. In machine learning, the Boruta algorithm is applied for feature selection. The selected features are subsequently utilized by XGBoost and random forest models for DR prevalence estimation. Use the Shapley additive explanations (SHAP) value to explain the independent contribution of eGDR. In the clinical cohort, we recruited patients who visited the Second Affiliated Hospital of Anhui Medical University from September 1, 2025, to December 30, 2025. A total of 297 participants who met the inclusion criteria were finally enrolled. Multivariable logistic regression and RCS curves were used to validate the findings from the NHANES analysis.
    Results: In the fully adjusted model, eGDR and self-reported DR prevalence show a significant negative linear correlation (OR = 0.79, 95% CI: 0.67-0.93, P = 0.0049). Subgroup and sensitivity analyses confirm the stability of this negative association. The Boruta algorithm identifies eGDR as a robust and important feature. Both the XGBoost (AUC = 0.773) and random forest (AUC = 0.764) models show moderate predictive performance, and eGDR has high variable importance. SHAP analysis indicates that eGDR, together with body mass index and income poverty, is a key determinant of self-reported DR prevalence. The results of the clinical cohort are like NHANES.
    Conclusion: This cross-sectional study suggested that lower eGDR is associated with a higher prevalence of self-reported DR. Accordingly, eGDR may serve as a potential marker for risk stratification rather than a causal or preventive factor. Prospective longitudinal research is necessary to confirm these findings and to explore whether a causal relationship exists.
    Keywords:  cross-sectional study; diabetic retinopathy; eGDR; insulin resistance; machine learning
    DOI:  https://doi.org/10.3389/fendo.2026.1853329
  22. Front Public Health. 2026 ;14 1850486
    Evaluation of large language model-generated information in diabetes health patient education: a scoping review.
    Shiya Hu, Binfen Lin, Fenglan Lin.
       Background: Diabetes is a leading cause of disability and death, posing a heavy healthcare burden globally. While standardized health education is crucial for glycemic control and mitigation of complications, traditional educational models face challenges due to insufficient scalability. The ongoing development of AI-based large language model (LLM) methods and technologies presents significant opportunities for health education in the field of diabetes.
    Objective: A scoping review of research on LLM-generated information for diabetes patient health education: Synthesizing current application status and performance outcomes.
    Methods: The Joanna Briggs Institute (JBI) evidence-based healthcare centre's scoping review guidance was utilized as the methodological framework, then five databases (PubMed, Embase, Web of Science, (American Psychological Association) APA PsycNet, and The Cochrane Library) were searched to retrieve studies from their inception to March 26, 2026. Two reviewers independently performed literature screening, full-text reading, and data extraction.
    Results: A total of 21 studies from nine countries were included. Application scenarios were categorized into five domains: general health education, dietary education, complication education, exercise education and technology education. Overall, the existing evidence indicates that LLMs perform well in terms of accuracy and completeness; however, significant limitations remain in readability, reliability, and usability. Moreover, ethical and safety concerns are prominent, including data security, fairness, patient safety, and liability.
    Conclusion: Despite existing technical and ethical challenges, LLMs still have potential as an auxiliary tool in diabetes health education. Future research needs to enhance technical design optimization, develop patient-centered designs, standardize evaluation metrics, and structured ethical oversight to further validate their practical application effects in diabetes health education for patients with diabetes.
    Keywords:  artificial intelligence; diabetes; education; large language model; scoping review
    DOI:  https://doi.org/10.3389/fpubh.2026.1850486
  23. Curr Diab Rep. 2026 Jul 10. pii: 20. [Epub ahead of print]26(1):
    Digital Management of Early-Onset Type 2 Diabetes: Empowerment, Challenges, and Future Outlook.
    Li-Jun Guo, Kah-Seng Low, Sok-Kun Tae, Lee-Ling Lim.
       PURPOSE OF REVIEW: Early-onset type 2 diabetes (EOT2D), defined as a diabetes diagnosis before 40 years of age, is rising globally and associated with an aggressive disease course and early complications. This review examines the role of digital health technologies (DHT) in addressing the unique clinical and life-course challenges of EOT2D.
    RECENT FINDINGS: DHT, including continuous glucose monitoring, mobile health applications, digital therapeutics, telemedicine, remote patient monitoring, wearable devices, and artificial intelligence-based analytics, have demonstrated modest improvements in glycemic control, weight management, and patient engagement in people with type 2 diabetes. However, evidence in adults with EOT2D remains limited. Compared with usual-onset T2D, people with EOT2D may derive particular benefits due to higher digital literacy, greater lifestyle variability, and longer anticipated disease duration. Although DHT shows promise for improving empowerment and care integration in EOT2D, important gaps persist, including a lack of EOT2D-specific trials, digital divide-related inequities, interoperability challenges, and reimbursement barriers. Future research should prioritize tailored interventions and hybrid care models to optimize long-term outcomes in this high-risk population.
    Keywords:  Artificial Intelligence; Continuous Glucose Monitoring; Digital Health; Digital Therapeutics; Early-Onset Type 2 Diabetes; Empowerment
    DOI:  https://doi.org/10.1007/s11892-026-01633-6
  24. JMIR Res Protoc. 2026 Jul 06. 15 e91699
    Artificial Intelligence-Assisted Screening for Patients With Diabetic Retinopathy and Age-Related Macular Degeneration in Family Medicine and Geriatric and Gerontology Care: Protocol for a Pragmatic Randomized Clinical Trial.
    Collaborators of AI Ophthalmology Research Group
       Background: Diabetic retinopathy (DR) and age-related macular degeneration (AMD) are 2 of the leading causes of vision loss worldwide. As population aging and diabetes prevalence increase, timely detection of these conditions has become essential. However, limited professionalism and insufficient training in ophthalmic screening among general medicine physicians may lead to delayed diagnosis and treatment. Artificial intelligence (AI)-assisted diagnostic tools may help to improve the screening of DR and AMD in routine clinical practice.
    Objective: This study aims to evaluate the clinical effectiveness and cost-effectiveness of AI-assisted fundus imaging for DR and AMD screening in adults with diabetes and older adults at risk of macular degeneration.
    Methods: This multicenter, 2-arm, parallel-group, open-label, individual-level randomized controlled trial and patient recruitment are performed at the settings of Family Medicine and Geriatric and Gerontology Care over 4 medical centers in Taiwan. Eligibility includes (1) diabetic individuals aged ≥20 years for DR screening, and (2) individuals aged ≥50 years for AMD screening. The study protocol has been approved by the ethics committees of all participating hospitals, and all participants will provide written informed consent.
    Results: The study was funded in September 2024, began on October 2, 2025, and is expected to be completed in December 2027. After the pilot implementation phase without randomization, participants will be randomized 1:1 into two groups: (1) AI-assisted screening, and (2) usual physician-only screening. The primary outcomes will include the detection rates (defined as participants with confirmed DR or AMD among all screened participants) and the positive predictive values (defined as participants with confirmed DR or AMD among those who tested positive). Cost-effectiveness analyses will be performed using data derived from the trial results.
    Conclusions: This study will provide robust evidence on the effectiveness of AI-assisted ophthalmic screening in improving patient eye health outcomes through timely screening and accurate early detection. This strategy may be cost-effective.
    Keywords:  age-related macular degeneration; artificial intelligence; cost-effectiveness; diabetic retinopathy; fundus photography; randomized controlled trial; screening
    DOI:  https://doi.org/10.2196/91699
This page is being maintained by Thomas Krichel. It was last updated on 2026‒07‒12 at 0:14.