bims-aukdir Biomed News
on Automated knowledge discovery in diabetes research
Issue of 2026–01–25
eighteen papers selected by
Mott Given
  1. An online interpretable machine learning model for predicting cardiometabolic multimorbidity risk in patients with type 2 diabetes mellitus.
  2. Continuous Glucose Monitoring-Based Machine Learning Identification of Diurnal Glycemic Patterns and Diabetes Distress in Type 2 Diabetes.
  3. Hemorrhage Segmentation in Fundus Images Using the U-Net 3+ Model: Performance Comparison Across Retinal Regions.
  4. Quantum denoising autoencoder improves retinal fundus image quality for early diabetic retinopathy screening.
  5. Hybrid Deep Learning Framework for Continuous Blood Glucose Monitoring and Gestational Diabetes Risk Prediction.
  6. Explainable machine learning model of coronary artery disease combined with diabetes: development and validation study.
  7. Combining thermography and artificial intelligence in comparison with a diabetic foot nurse for diabetic foot ulcer detection: A diagnostic accuracy study.
  8. Machine learning-based prediction of diabetic retinopathy from pupillary abnormalities in a South Indian population.
  9. SERS-based deep learning approach for early detection of gestational diabetes mellitus.
  10. Machine learning-based online prediction of nocturnal hypoglycemia in elderly patients with type 2 diabetes.
  11. First trimester prediction of gestational diabetes mellitus by machine learning in twin pregnancies.
  12. Machine learning-based prediction model for 30-day readmission risk in elderly patients with type 2 diabetes mellitus and heart failure: a retrospective cohort study with SHAP interpretability analysis.
  13. Efficacy of an AI-Enabled Low Glucose Prediction: A Pooled Performance Analysis With Capillary Blood Glucose as Ground Truth.
  14. A risk model to predict atrial fibrillation in diabetes using machine learning: The ACCORD study.
  15. Deep exploration and precise identification of key risk factors for diabetic peripheral neuropathy using the random forest algorithm.
  16. Lifespan Pancreas Morphology for Control Versus Type 2 Diabetes Using AI on Largescale Clinical Imaging.
  17. Diagnostic Accuracy of Artificial Intelligence-Enabled Retinal Biomarkers for Detecting Type 2 Diabetes and Prediabetes Among Asian Indians [DART Study].
  18. Randomized trial and multi-omics, machine learning-based mechanistic exploration of daixie decoction granules in type 2 diabetes.
  1. Sci Rep. 2026 Jan 21.
    An online interpretable machine learning model for predicting cardiometabolic multimorbidity risk in patients with type 2 diabetes mellitus.
    Xiaohan Liu, Cheng Li, Xiaotong Huo, Junjiao Liu, Jie Liu, Wenjun Cao, Jianzhong Zheng.
      Cardiometabolic multimorbidity (CMM), a major complication in type 2 diabetes mellitus (T2DM), increases mortality and healthcare burden. Early identification of high-risk individuals is crucial for precision intervention. This study aimed to develop and validate an online interpretable machine learning system for forecasting the CMM risk in T2DM populations to facilitate personalized decision-making and early intervention. We used data from 793 T2DM patients from a tertiary hospital in Shanxi Province as the derivation cohort, divided into training (80%) and internal validation (20%) sets, with 360 cases from another independent center for external validation. Feature selection was performed through recursive feature elimination with random forest algorithm. We employed six machine learning algorithms to develop the CMM risk model. Model performance was evaluated using accuracy, precision, recall, F1-score, and area under the curve (AUC). The SHapley Additive exPlanations (SHAP) and Local Interpretable Model-Agnostic Explanations (LIME) provided model interpretability. After feature screening, nine predictors were included in the model. In internal validation, the Stacking model achieved the highest AUC (0.868), maintaining good external validation performance with an AUC of 0.822. The web-based system was accessible on https://t2dmcmmpredictionweb.streamlit.app/. This system assisted healthcare providers to identify high-risk populations early and facilitate timely intervention to mitigate disease progression.
    Keywords:  Cardiometabolic multimorbidity; Machine learning; Prediction model; Type 2 diabetes mellitus
    DOI:  https://doi.org/10.1038/s41598-026-36923-2
  2. J Diabetes Sci Technol. 2026 Jan 18. 19322968251412449
    Continuous Glucose Monitoring-Based Machine Learning Identification of Diurnal Glycemic Patterns and Diabetes Distress in Type 2 Diabetes.
    Minjung Lee, Soohyun Nam.
       BACKGROUND: To identify diurnal glycemic patterns in adults with type 2 diabetes (T2D) using continuous glucose monitoring (CGM)-based machine learning and examine their association with diabetes distress, a key psychosocial outcome.
    METHODS: In this observational study, 137 adults with T2D wore blinded CGM (FreeStyle Libre Pro), yielding 1657 days of data. Glycemic patterns were identified using unsupervised machine learning via Gaussian mixture modeling, validated with Bayesian information criterion and silhouette scores. Diabetes distress was assessed with the 17-item Diabetes Distress Scale and analyzed through analysis of covariance (ANCOVA), adjusting for age, sex, body mass index, diabetes duration, and glucose management indicator.
    RESULTS: Clustering identified four distinct glycemic profiles: Cluster 1 (suboptimal control, nocturnal hypoglycemia; 15.8%), Cluster 2 (suboptimal control, nocturnal hyperglycemia; 27.1%), Cluster 3 (poorly controlled, prolonged hyperglycemia; 21.1%), and Cluster 4 (well controlled; 36.1%). Diabetes distress scores varied significantly: participants in Cluster 3 reported the highest distress (mean = 2.37, 95% CI = 1.99-2.76), while Cluster 4 reported the lowest (mean = 1.67, 95% CI = 1.48-1.86; P = .03). Effect sizes indicated differences corresponded to clinically meaningful categories of "little or no distress" vs "moderate distress."
    CONCLUSIONS: CGM-based machine learning identified physiologically distinct glycemic phenotypes that were also associated with psychosocial burden. This work demonstrates the added value of integrating CGM-derived profiles with patient-reported outcomes. These findings highlight the potential of CGM phenotyping to support precision diabetes care by enabling early identification of high-risk subgroups, guiding tailored behavioral and psychosocial interventions, and informing technology-enabled decision tools that connect physiological monitoring with emotional well-being in T2D management.
    Keywords:  CGM; clustering; continuous glucose monitoring; diabetes distress; diurnal glycemic patterns; machine learning; patient-reported outcomes; precision medicine; type 2 diabetes
    DOI:  https://doi.org/10.1177/19322968251412449
  3. J Imaging Inform Med. 2026 Jan 21.
    Hemorrhage Segmentation in Fundus Images Using the U-Net 3+ Model: Performance Comparison Across Retinal Regions.
    Yeong Hun Kang, Young Jae Kim, Kwang Gi Kim.
      Diabetic retinopathy (DR) is one of the most common complications of diabetes, and timely detection of retinal hemorrhages is essential for preventing vision loss. This study evaluates the U-Net3 + model for pixel-level hemorrhage segmentation in fundus images and examines its performance across clinically meaningful retinal regions. Model performance was assessed using accuracy, sensitivity, specificity, and Dice score and further analyzed across perivascular and extravascular areas, perifoveal and extrafoveal regions, fovea-centered quadrants, and images stratified by hemorrhage burden. U-Net3 + achieved strong overall performance, with 99.93% accuracy, 87.03% sensitivity, 99.97% specificity, and an 85.02% Dice score. Higher segmentation accuracy was observed in extravascular regions and within the foveal area, while quadrant-wise performance remained largely consistent. Images with greater hemorrhage burden demonstrated higher segmentation reliability. These findings highlight the importance of region-aware evaluation and suggest that U-Net3 + can provide clinically meaningful support for automated DR screening. Further validation using larger and multi-center datasets is required to enhance the model's generalizability for real-world clinical deployment.
    Keywords:  Deep learning models; Diabetic retinopathy; Hemorrhage segmentation; Image analysis
    DOI:  https://doi.org/10.1007/s10278-025-01837-4
  4. Sci Rep. 2026 Jan 21.
    Quantum denoising autoencoder improves retinal fundus image quality for early diabetic retinopathy screening.
    Rajitha Chilukuri, Praveen P, Ranjith Kumar Gatla, Reem A Almenweer.
      Diabetic Retinopathy (DR) is a critical source of blindness that can be prevented globally, and accurate analysis of retinal fundus images enables early detection. Fundus images are often affected by multiple noise sources, which impair image quality and hinder the observation of delicate retinal structures, including microaneurysms and small blood vessels. Deep learning driven denoising models are computationally intensive and prone to overfitting on small medical datasets. In order to overcome these shortcomings, the present paper suggests a Quantum Denoising Autoencoder (QDAE), a hybrid quantum-classical architecture, which uses convolutional feature coding with parameterized quantum circuits (PQCs) in latent space. The suggested QDAE applies quantum superposition and entanglement to improve the latent representations, thereby improving denoising and retinal detail preservation. Experiments on the Diabetic Retinopathy 224 × 224 (2019) dataset show that QDAE performs considerably better than classical denoising architectures, including CAE, ResNet, and DnCNN with PSNR of 38.8 dB, SSIM of 0.96, and AMI of 0.88. The approach preserves delicate retinal patterns and intensity consistency, while incurring a slight computational overhead associated with shallow quantum circuits. The results presented above demonstrate that QDAE is a potential quantum-aided architecture for denoising retinal images and a feasible preprocessing procedure in early diabetic retinopathy.
    Keywords:  Deep learning; Diabetic retinopathy; Fundus image denoising; Medical image processing; Parameterized quantum circuits (PQC); Quantum computing
    DOI:  https://doi.org/10.1038/s41598-026-35540-3
  5. Curr Diabetes Rev. 2026 Jan 09.
    Hybrid Deep Learning Framework for Continuous Blood Glucose Monitoring and Gestational Diabetes Risk Prediction.
    Kiran Kumar Chanumolu, Muni Nagamani G, Muni Hemalatha G.
       BACKGROUND: Gestational diabetes mellitus (GDM) affects almost 10%-12% of pregnancies worldwide, threatening maternal and fetal life. Continuous glucose monitoring (CGM) forms the backbone of managing GDM, and the current methodologies largely disregard physiological and behavioral factors, thereby greatly reducing accuracy and clinical interpretability.
    METHODS: A hybrid deep learning framework was developed by fusing CGM with multi-sensing modality data, including heart rate, activity levels, sleep patterns, and dietary intake. For data preprocessing, Kalman filtering was applied for temporal alignment, adaptive normalization provided outlier handling and imputation, while the CNN-BiLSTM backbone with attention was harnessed for feature extraction. A Multi-Task Attention Fusion Network (MTAFN) was used to predict glucose values and classify GDM risk simultaneously, while SHAP and dynamic smoothing contributed to interpretability sets.
    RESULTS: The framework was validated on an extended OhioT1DM dataset with adaptations for pregnancy. It reached a glucose prediction RMSE of 9.8 mg/dL and a GDM risk classification accuracy of 93%. Compared to competitive approaches, the present solution attained a 25% better accuracy on interpretability and an improvement in sensitivity and specificity of about 4-6% across various physiological conditions.
    DISCUSSION: The use of multi-sensing data increased prediction robustness by capturing complex physiological dependencies. The SHAP-based interpretability justified the predictions through a physiological lens. With an attention mechanism for feature weighting, it was possible to identify crucial variables like meal intake and nighttime variability in the workflow sets.
    CONCLUSION: The hybrid framework proposed here is reliable for clinically interpretable continuous glucose monitoring and GDM risk predictions. Its application with high reliability can lead to integrating it within clinical protocols for real-time maternal care sets.
    Keywords:  Continuous glucose monitoring; deep learning; gestational diabetes prediction; interpretive dynamic smoothing; multi-sensor data integration; scenarios.
    DOI:  https://doi.org/10.2174/0115733998380389251111041618
  6. Front Cardiovasc Med. 2025 ;12 1674287
    Explainable machine learning model of coronary artery disease combined with diabetes: development and validation study.
    Xujie Wang, Shipeng Wang, Xuhui Liu, Rongfei Xie, Shasha Liang, Biyun Wang, Yuke Zhang.
       Background: Coronary artery disease (CAD) demonstrates a strong bidirectional association with diabetes mellitus, which not only elevates cardiovascular disease risk but also correlates with poorer clinical prognosis. Prognostication in patients with comorbid CAD and diabetes remains a critical clinical challenge, significantly influencing therapeutic decision-making. Leveraging readily available clinical parameters for predicting adverse outcomes in this population offers substantial clinical value. This investigation employs machine learning algorithms to develop predictive models for prognostic assessment in CAD patients with diabetes comorbidity.
    Method: We conducted a retrospective cohort study of 389 patients with comorbid coronary artery disease (CAD) and diabetes mellitus. The cohort was randomly allocated into a training set (n = 273) and an internal validation set (n = 116). Feature selection utilized LASSO regression followed by backward stepwise Cox regression analysis. A nomogram incorporating independent predictors was developed for clinical application. Model performance was assessed through discrimination metrics, calibration plots, and decision curve analysis (DCA). Random survival forest analysis validated the clinical significance of selected variables.
    Result: Our modeling approach employed a systematic methodology: LASSO regression for initial feature selection followed by backward stepwise Cox regression analysis, yielding eight independent predictors.The final model incorporated hemoglobin, INR, albumin, NT-proBNP, age, fibrinogen, diuretic use, and digitalis therapy. The integrated model demonstrated strong discriminative performance for mortality prediction across both training (AUC = 0.846, 0.838, 0.82) and validation cohorts (AUC = 0.824, 0.813, 0.798) at 3-, 5-, and 8-year intervals. Calibration plots and decision curve analysis confirmed model reliability and clinical utility over time. A nomogram was developed to facilitate individualized risk stratification. Kaplan-Meier analysis showed significant survival stratification by individual predictors, and restricted cubic spline analysis identified non-linear associations between continuous variables and mortality. Random survival forest analysis prioritized five key predictors (hemoglobin, INR, albumin, NT-proBNP, age). Comparative evaluation against the 9-variable model confirmed superior performance of the comprehensive model across all timepoints.
    Conclusion: Our multimodal prognostic model demonstrated robust performance in predicting all-cause mortality among patients with coronary artery disease and diabetes comorbidity. The nomogram's capacity for personalized risk estimation offers potential utility in clinical decision-making and patient stratification.
    Keywords:  coronary artery disease; diabetes mellitus; machine learning; mortality; prognostic model
    DOI:  https://doi.org/10.3389/fcvm.2025.1674287
  7. Digit Health. 2026 Jan-Dec;12:12 20552076261416807
    Combining thermography and artificial intelligence in comparison with a diabetic foot nurse for diabetic foot ulcer detection: A diagnostic accuracy study.
    Khansa Shara, Mustafa Alghali, Waseem Abu-Ashour, Ahmad T Almnaizel, Tamara Sunbul, Nada Baatiah, Kariman Attal, Ibtihal Al Attallah, Baneen Sawad, Meshari Alwashmi.
       Background: Early detection of diabetic foot complications is essential to prevent ulcers and amputations. Thermographic imaging offers a non-invasive method for identifying risk, but clinical interpretation has traditionally relied on human thermographers. Artificial intelligence (AI) may offer a more scalable and objective alternative.
    Objective: To evaluate the diagnostic performance of an AI-powered thermographic screening tool in identifying risk for diabetic foot complications, compared to nurse-led clinical assessment.
    Methods: We conducted a cross-sectional study of 100 adults with diabetes undergoing routine foot screening. For each participant, a smartphone-based thermal imaging device was first used to capture plantar images, from which the AI model generated risk scores (0-3). Second, a diabetic foot nurse performed a clinical examination and assigned the reference risk scores (0-3). Absolute temperature differences were computed from thermal images, and diagnostic accuracy metrics were calculated using the nurse assessment as the reference standard.
    Results: The AI system demonstrated 100% sensitivity, 96.8% specificity, 66.7% positive predictive value, and 100% negative predictive value for detecting moderate-to-high risk cases. There was a strong correlation between AI and nurse scores (ρ = 0.973), and both assessors showed increasing temperature asymmetry with higher risk levels.
    Conclusions: The AI model accurately detected all moderate-to-high risk cases flagged by the nurse, with high sensitivity and specificity. Its strong alignment with thermal data and consistent scoring suggest its value as a scalable and reproducible adjunct for diabetic foot screening. Further validation in longitudinal settings may support broader integration in remote and primary care environments.
    Keywords:  Diabetic foot ulcers; artificial intelligence; diagnostic accuracy; early detection; screening; thermography
    DOI:  https://doi.org/10.1177/20552076261416807
  8. PLoS One. 2026 ;21(1): e0340802
    Machine learning-based prediction of diabetic retinopathy from pupillary abnormalities in a South Indian population.
    Janani Surya, S Tamilselvi, Maitreyee Roy, Sivaraj Chinnasamy, Rajiv Raman, M Suchetha, Ganesh Rajendran.
      Diabetic retinopathy (DR) is a common complication of diabetes that can lead to vision loss. Early detection and prevention of DR is crucial to reduce the burden of this disease. The purpose of this study was to build a prediction model for DR using pupillary abnormalities as biomarkers. Pupillary parameters including Dark-adapted Baseline Pupillary Diameter (BPD), Amplitude of Pupillary Constriction (APC), Velocity of Pupillary Constriction (VPC), Amplitude of Pupil Re-dilatation after Maximum Constriction, and Velocity of Pupillary Dilatation (VPD) were collected and analyzed using machine learning algorithm including Support Vector Machine, Decision Trees, Artificial Neural Networks (ANN), Logistic Regressions, Random Forest, Naive Bayes Classifier. Utilizing ROC analysis and the Youden index, this study identified cut-off values for pupillary abnormalities to detect DR risk. The study found that ANN performed well with an accuracy of 0.807 (95% CI: 0.65-0.94) and AUC of 0.879 (95% CI: 0.71-0.98) in predicting DR using pupillary abnormalities as biomarkers. The findings of this research offer significant insights into the predictive value of pupillary abnormalities for DR, establishing a strong foundation for early intervention strategies. Particularly, the superior performance of ANN in detecting DR presents an opportunity to refine risk stratification and prevention approaches, potentially transforming the prognosis for individuals at elevated risk of this condition.
    DOI:  https://doi.org/10.1371/journal.pone.0340802
  9. Spectrochim Acta A Mol Biomol Spectrosc. 2026 Jan 13. pii: S1386-1425(26)00043-0. [Epub ahead of print]351 127472
    SERS-based deep learning approach for early detection of gestational diabetes mellitus.
    Huizhen Lin, Jiawang Chen, Yiming Chen, Dechan Lu.
      Early and precise diagnosis of gestational diabetes mellitus (GDM) is crucial for improving maternal and neonatal outcomes and reducing the risk of adverse pregnancy events. However, current clinical screening methods for GDM still exhibit limitations in detection speed, sensitivity and convenience, making it difficult to meet the clinical demand for rapid early-pregnancy screening. To address this, we propose a novel strategy for early GDM diagnosis based on surface-enhanced Raman spectroscopy (SERS) combined with deep learning, aiming to achieve rapid and accurate early screening. Characteristic SERS spectra of serum were obtained using a substrate based on silver nanoparticles (Ag NPs). A fused PCA-CNN model integrating principal component analysis (PCA) for dimensionality reduction and a one-dimensional convolutional neural network (1D-CNN) for feature learning was developed. The PCA-CNN model effectively extracts potential biomarker features from serum SERS spectra, achieving a diagnostic accuracy of 93.7%, with sensitivity and specificity of 0.95 and 0.93, respectively. Moreover, the entire detection process can be completed within 30 min, requires about 2.5 μL of serum per sample, and involves minimal preprocessing, highlighting both efficiency and practicality. This study provides a novel method for early GDM screening that combines high diagnostic performance with clinical applicability, offering promising technical support for early intervention and clinical management of GDM.
    Keywords:  Gestational diabetes mellitus; Label free; PCA-CNN; SERS; Serum
    DOI:  https://doi.org/10.1016/j.saa.2026.127472
  10. Front Endocrinol (Lausanne). 2025 ;16 1685969
    Machine learning-based online prediction of nocturnal hypoglycemia in elderly patients with type 2 diabetes.
    Yuntong Liu, Chenhua Guo, Xinyu Li, Shen Li, Jiajun Huang, Liang Zhao, Yan Zhu, Xuhan Liu, Bing Wang, Rui Lin, Jingshi Wang, Zhengnan Gao, Jing Gao, Yingshu Liu.
       Context: Nocturnal hypoglycemia (NH) is a common adverse event in elderly patients with type 2 diabetes (T2D). This study aims to develop a clinically applicable model for predicting the risk of NH in elderly patients with T2D.
    Methods: This retrospective cohort study, conducted from May 2018 to June 2024, analyzed 1,128 elderly T2D patients undergoing continuous glucose monitoring, with an independent validation involving 100 outpatients. Clinical characteristics were collected, and feature engineering was performed to select a manageable set of clinically accessible features. An ensemble model was developed using multiple base models and a stacking approach. The best-performing model was deployed as an online risk calculator.
    Results: Of the development set, 288 (25.5%) experienced NH, while 40 (40%) of the independent validation cohort experienced NH. The final ensemble model, "RF-ET-KNN", combined random forest, Extra Trees, and K-nearest neighbor as base learners, with Extra Trees serving as the meta-learner. It incorporated eleven clinical features and achieved an AUROC of 0.926 and sensitivity of 0.853 on the test set, and an AUROC of 0.947 and sensitivity of 0.929 on the internal validation set. SHAP analysis identified that daytime lowest blood glucose (BG), fasting blood glucose (FBG), and daytime hypoglycemia events were closely related to NH. A user-friendly calculator is available at http://122.51.219.102:8000/.
    Conclusion: The "RF-ET-KNN" model, integrating eleven clinically accessible features, effectively predicts NH in elderly T2D patients. Daytime lowest BG, FBG, and daytime hypoglycemia events were significant risk factors.
    Keywords:  clinical prediction model; elderly people; ensemble learning; nocturnal hypoglycemia; type 2 diabetes
    DOI:  https://doi.org/10.3389/fendo.2025.1685969
  11. Arch Gynecol Obstet. 2026 Jan 20. 313(1): 52
    First trimester prediction of gestational diabetes mellitus by machine learning in twin pregnancies.
    Yoram Louzoun, Tamar Michelson, Mar Bennasar, Ran Svirsky, Elisa Bevilacqua, Nadav Kugler, Karl Kagan, Richard Nicholas Brown, Heidy Portillo Rodriguez, Anna Goncé, Antoni Borrell, Julia Ponce, Annegret Geipel, Adeline Walter, Corinna Simonini, Brigitte Strizek, Tanja Lennartz, Armin Bauer, Federica Meli, Eleonora Torcia, Adi Sharabi-Nov, Ron Maymon, Kypros H Nicolaides, Hamutal Meiri.
       INTRODUCTION: We aimed to develop a machine learning model for first-trimester prediction of gestational diabetes mellitus (GDM) in twin pregnancies using a prospective international, multi-center cohort and identify useful predictive markers.
    METHODS: Pregnant women with two live fetuses were enrolled at 11 + 0 to 13 + 6 weeks' gestation and followed until delivery. GDM was diagnosed at 24-28 weeks' gestation using the two-stage GCT and OGTT tests. Biochemical, biophysical, and blood assessments were conducted at three periods during pregnancy. Multiple machine learning models evaluated demographic, clinical, and laboratory parameters, including maternal factors (BMI, age, medical history), sonographic markers (crown rump length, estimated fetal weight, uterine artery pulsatility index), and blood and biochemical markers (placental growth factors, blood glucose, cell counts). LightGBM, XGBoost, and logistic regression models were compared using area under the curve (AUC) analysis.
    RESULTS: Among 596 women, 99 (16.6%) developed GDM. LightGBM demonstrated superior performance (AUC = 0.72, 95% CI 0.69-0.75). First-trimester high BMI was the strongest predictor, followed by elevated white blood cell counts and platelet levels. Detection rates (DR) were 28% and 42% at 10% and 20% false positive rates (FPR), respectively. Previous GDM was associated with an increased risk for GDM.
    DISCUSSION: GDM in twins is associated with certain characteristics of the first-trimester. Information from later trimesters has a limited impact. The GDM probability risk score increased with the severity of the treatment. An app to predict this score is available at: twin-pe.math.biu.ac.il.
    Keywords:  Machine learning; Prediction of GDM; Screening markers; Twin pregnancy
    DOI:  https://doi.org/10.1007/s00404-025-08262-6
  12. Front Cardiovasc Med. 2025 ;12 1673159
    Machine learning-based prediction model for 30-day readmission risk in elderly patients with type 2 diabetes mellitus and heart failure: a retrospective cohort study with SHAP interpretability analysis.
    Lei Wang, Chao Wei, Yue Chen, Da-Sheng Lu, Hong-Xiang Zhang, Mo Yang.
       Objective: Among elderly populations with concurrent type 2 diabetes mellitus (T2DM) and heart failure (HF), 30-day hospital readmission rates range 10%-25%. Conventional risk evaluation instruments show restricted predictive performance (AUC < 0.70) in this multimorbid group. This research aimed to construct and verify an artificial intelligence-based algorithm for assessing 30-day readmission probability in elderly T2DM-HF patients.
    Methods: This retrospective cohort study included 870 participants ≥65 years with T2DM and HF (January 2020-December 2023), randomly divided into training (n = 609, 70%) and validation (n = 261, 30%) cohorts. Variable selection utilized Least Absolute Shrinkage and Selection Operator with ten-fold cross-validation. Eight machine learning algorithms were evaluated: logistic regression, random forest, gradient boosting machines, support vector machines, neural networks, convolutional neural networks, AdaBoost, and stacking ensemble. Model interpretability was enhanced using SHapley Additive exPlanations analysis.
    Results: Overall 30-day readmission rate was 12.4% (108/870 patients). The Stacking Ensemble model achieved superior performance with AUC 0.867 (95% CI: 0.830-0.904), accuracy 79.4%, sensitivity 74.9%, and specificity 84.0%. Fourteen key predictors were identified, with C-reactive protein, estimated glomerular filtration rate, and B-type natriuretic peptide as most influential factors.
    Conclusion: This study developed a high-performing, interpretable machine learning model for predicting 30-day readmission risk, providing a valuable clinical decision-making tool.
    Keywords:  30-day readmission; heart failure; machine learning; risk prediction; type 2 diabetes mellitus
    DOI:  https://doi.org/10.3389/fcvm.2025.1673159
  13. J Diabetes Sci Technol. 2026 Jan 18. 19322968251412451
    Efficacy of an AI-Enabled Low Glucose Prediction: A Pooled Performance Analysis With Capillary Blood Glucose as Ground Truth.
    Timor Glatzer, Ajandek Peak, Eemeli Leppäaho, Patrick Lustenberger, Pau Herrero, Magí Andorrà, Ellen van Maren.
       BACKGROUND: Hypoglycemia is a critical challenge for insulin-dependent people with diabetes using multiple daily injections (MDI), who rely on reactive responses to continuous glucose monitoring (CGM) alerts. To meet the need for a proactive safety tool, we evaluated the performance of the Low Glucose Predict (LGP) feature in the Accu-Chek SmartGuide Predict App.
    METHODS: This retrospective analysis pooled data from three prospective trials, including 85 subjects over 2709 recording days. The LGP feature uses a XGBoost model to predict low glucose events up to 30 minutes in advance. Performance was assessed rigorously against both capillary blood glucose (BG) and CGM values, including an analysis with "close-call" predictions (+10 mg/dL above the threshold). Metrics included sensitivity, specificity, and ROC-AUC.
    RESULTS: Against the stringent capillary BG reference, LGP showed high performance: sensitivity of 87.13% and specificity of 97.43% (ROC-AUC 0.9787). Including close-call events improved sensitivity to 91.89% and specificity to 98.09%. Referenced against CGM, sensitivity was 94.40% and specificity was 98.25%. The system provided an actionable mean lead time of 14.71 ± 8.30 minutes (CGM reference), with a low average daily true notification rate of 1.31 (2.60 including close-calls).
    CONCLUSION: The LGP feature is an accurate, highly sensitive, and specific tool for timely, proactive low glucose prediction, validated against both capillary BG and CGM. This predictive intelligence is a crucial mechanism for people with diabetes to safely mitigate hypoglycemia risk, addressing a significant clinical gap and potentially reducing fear of hypoglycemia and diabetes distress.
    Keywords:  alarm fatigue; artificial intelligence; continuous glucose monitoring (CGM); hypoglycemia prediction
    DOI:  https://doi.org/10.1177/19322968251412451
  14. Am Heart J Plus. 2026 Feb;62 100715
    A risk model to predict atrial fibrillation in diabetes using machine learning: The ACCORD study.
    Erik J Offerman, Joseph Phan, Sarah Harirforoosh, Wenjun Fan, Nathan D Wong, David M Donaldson.
       Background: Machine learning (ML) may improve prediction of atrial fibrillation (AF), but its value compared with traditional models such as Cohorts for Heart and Aging Research in Genomic Epidemiology (CHARGE-AF) in patients with diabetes remains unclear.
    Methods: Among 9,307 patients in the Action to Control Cardiovascular Risk in Diabetes (ACCORD) with type 2 diabetes and no prior AF, a random forest (RF) classifier using clinical and metabolic variables was compared with a CHARGE-AF Cox model. Discrimination was assessed by five-fold cross-validated area under receiver operating curve (AUC).
    Results: Over 6.26 years, 175 patients developed AF. The RF model (AUC = 0.731) performed comparably to CHARGE-AF (AUC = 0.756; p = 0.18). Age, waist circumference, race, total cholesterol, and estimated glomerular filtration rate were the top predictors.
    Conclusion: ML matched CHARGE-AF performance and revealed distinct predictors supporting personalized AF risk prevention.
    Keywords:  Atrial fibrillation; Cardiovascular epidemiology; Diabetes mellitus; Machine learning; Random forest; Risk prediction
    DOI:  https://doi.org/10.1016/j.ahjo.2026.100715
  15. Front Endocrinol (Lausanne). 2025 ;16 1740545
    Deep exploration and precise identification of key risk factors for diabetic peripheral neuropathy using the random forest algorithm.
    Yongnan Li, Yongsheng Li, Gan Sen, Bin Hu.
       Background: Diabetic peripheral neuropathy (DPN) is a prevalent and highly disabling complication of diabetes mellitus, associated with markedly increased rates of disability and mortality. Timely intervention and effective management have been consistently shown to substantially reduce the risk of DPN onset and progression.
    Methods: This retrospective cohort study analyzed 1, 004 hospitalized patients with type 2 diabetes mellitus (T2DM) admitted to the endocrinology department of a hospital in Jiangsu Province, China. A risk prediction model for DPN was developed using the Random Forest (RF) algorithm, while logistic regression analysis was employed to identify the major risk factors. The overarching aim was to provide a reliable risk assessment tool for clinical application.
    Findings: Five principal factors were identified as significantly associated with DPN risk: age (OR = 1.257, 95% CI [1.188-1.367], p < 0.001), serum 25(OH)D3 levels (OR = 0.791, 95% CI [0.759-0.854], p < 0.001), duration of diabetes (OR = 1.431, 95% CI [1.285-1.617], p < 0.001), glycated hemoglobin (HbA1c) (OR = 1.236, 95% CI [1.197-1.391], p < 0.001), and glycated serum protein (GSP) (OR = 1.091, 95% CI [1.047-1.201], p = 0.017). A DPN risk prediction model incorporating these variables achieved an area under the receiver operating characteristic curve (AUC) of 0.829 (95% CI: 0.802-0.857), demonstrating excellent discriminatory performance.
    Interpretation: The Random Forest-based DPN risk prediction model successfully identified five critical risk factors, offering a solid theoretical foundation for personalized strategies in DPN prevention and management among patients with diabetes. This model exhibits high predictive utility in clinical practice.
    Keywords:  25(OH)D3; GPS; diabetic peripheral neuropathy; random forest; type 2diabetes mellitus (T2DM)
    DOI:  https://doi.org/10.3389/fendo.2025.1740545
  16. Clin Anat. 2026 Jan 20.
    Lifespan Pancreas Morphology for Control Versus Type 2 Diabetes Using AI on Largescale Clinical Imaging.
    Lucas W Remedios, Chloe Cho, Trent M Schwartz, Dingjie Su, Gaurav Rudravaram, Chenyu Gao, Aravind R Krishnan, Adam M Saunders, Michael E Kim, Shunxing Bao, Thomas A Lasko, Alvin C Powers, Bennett A Landman, John Virostko.
      Understanding how pancreas size and shape change with normal aging is critical for establishing a baseline to detect deviations in type 2 diabetes and other pancreatic disease. We measure pancreas size and shape using morphological measurements from early development through aging (ages 0-90). Our goals are to (1) identify reliable clinical imaging modalities for artificial intelligence (AI) based pancreas measurement, (2) establish normative morphological aging trends, and (3) detect potential deviations in type 2 diabetes. We analyzed a clinically acquired dataset of 2533 patients imaged with abdominal computed tomography (CT) or magnetic resonance imaging (MRI). The patients did not have cancer, pancreas pathology, sepsis, or trauma. We resampled the scans to 3 mm isotropic resolution, segmented the pancreas using automated methods, and extracted 13 morphological pancreas features across the lifespan. First, we assessed pancreas volume trajectories in 1858 control patients across contrast CT, non-contrast CT, and MRI to determine which modalities provide consistent lifespan trends. Second, we characterized distributions of normative morphological patterns stratified by age group and sex. Third, we used covariate-adjusted generative additive models for location, scale, and shape (GAMLSS) regression to model pancreas morphology trends in 1350 patients matched for age, sex, and type 2 diabetes status to identify any deviations from normative aging associated with type 2 diabetes. We selected CT for the main analyses of this study, since the MRI appeared to yield different pancreas measurements than CT using our AI-based method on this dataset of clinically acquired scans. When adjusting for confounders, the aging trends for 10 of 13 morphological features were significantly different between patients with type 2 diabetes and non-diabetic controls (p < 0.05 after multiple comparisons corrections). Additionally, we characterized normative morphological aging trends of the pancreas across 13 morphological measurements. We provide lifespan trends demonstrating that the size and shape of the pancreas are altered in type 2 diabetes using 675 control patients and 675 diabetes patients. Moreover, our findings reinforce that the pancreas is smaller in type 2 diabetes. Additionally, we contribute a reference of lifespan pancreas morphology from a large cohort of non-diabetic control patients in a clinical setting.
    Keywords:  CT; MRI; aging; multimodal; pancreas; shape; volume
    DOI:  https://doi.org/10.1002/ca.70077
  17. Diabetes Technol Ther. 2026 Jan 23. 15209156261417296
    Diagnostic Accuracy of Artificial Intelligence-Enabled Retinal Biomarkers for Detecting Type 2 Diabetes and Prediabetes Among Asian Indians [DART Study].
    Soujanya Kaup, Sudeshna Sil Kar, Rodrigo M Carrillo-Larco, Rohan Dhamdhere, Mendel Lebowitz, Ranjit Mohan Anjana, Ram Jagannathan, Mohammed K Ali, Viswanathan Mohan, K M Venkat Narayan, Anant Madabhushi, Ramachandran Rajalakshmi.
       PURPOSE: To identify retinal microvascular features that distinguish normoglycemia from diabetes mellitus (DM)[Study-1] and prediabetes (PreDM)[Study-2] among Asian Indians using artificial intelligence (AI); evaluate their diagnostic accuracy; and examine independent associations of oculomics scores [Oculomic Diabetes Score (ODiS) and Prediabetes Score (OPreS)] with glycemic status.
    METHODS: We analyzed 273 retinal images from 139 participants (19 = normoglycemia, 100 = DM, 20 = PreDM; mean age: 49.4 ± 12.9 yrs; male: 71.2%) with dataset randomly split (50:50) into training and test sets. We extracted 226 quantitative vessel tortuosity features separately for arteries and veins using machine vision-based approaches. Top six discriminating features were separately selected (using Wilcoxon Rank-Sum test and Linear Discriminant Analysis) on training sets and validated on independent blinded test sets. Model performances were evaluated using area under precision-recall curve (AUPRC) for DM and area under the receiver operating characteristic curve (AUC) for PreDM. Independent association of ODiS and OPreS (adjusting for age, sex, height, weight, blood pressure, serum cholesterol, serum creatinine) was assessed by multivariable logistic regression.
    RESULTS: Specific oculomics-based retinal vascular features distinguished DM/PreDM from normoglycemia. Vein-only model achieved AUPRC = 0.96 (95% confidence interval [95% CI]: 0.9-1.00) (sensitivity = 95%, specificity = 72.2%, precision = 95%) on Sv1 for DM and an AUC = 0.80 (95% CI: 0.63-0.94) (sensitivity = 70%, specificity = 80%, precision = 82.35%) on Sv2 for PreDM. Oculomics scores were independently associated with DM(ODiS) [Adjusted Odds ratio (AdjOR): 2.00 (95% CI: 1.56-2.57, P < 0.001)] but not with PreDM(OPreS) [AdjOR: 1.32 (95% CI: 0.65-2.71, P = 0.48)].
    CONCLUSIONS: In this proof-of-concept study, AI-informed retinal venous features on routine fundus images, with further prospective and multisite validation, could potentially serve as noninvasive DM detection using AI models among Asian Indians.
    Keywords:  artificial intelligence; color fundus images; diabetes; oculomics; prediabetes; retinal biomarkers
    DOI:  https://doi.org/10.1177/15209156261417296
  18. Front Pharmacol. 2025 ;16 1723584
    Randomized trial and multi-omics, machine learning-based mechanistic exploration of daixie decoction granules in type 2 diabetes.
    Zhong Zheng, Kepei Zhang, Xiaogang Ma, Yuezhou Qian, Miao Wang.
       Introduction: Traditional Chinese Medicine (TCM) offers multi-target strategies for Type 2 Diabetes Mellitus (T2DM), but its mechanisms are unclear. This study combined a randomized controlled trial (RCT) with a multi-omics approach to evaluate the efficacy of Daixie Decoction granules (DDG) as an add-on therapy to metformin and to generate mechanistic hypotheses using a multi-omics framework.
    Methods: We conducted a randomized, double-blind, placebo-controlled trial involving 136 randomized and 128 completed with DDG plus metformin or placebo plus metformin for 6 months. Mechanistic prediction was based on network pharmacology, integration of T2DM-related genes from public databases (GeneCards, DisGeNET, OMIM), and transcriptomic differentially expressed genes (DEGs) from GEO. Seven machine learning algorithms were applied to prioritize core targets from the overlapping candidate list. A nested serum proteomics sub-study within the randomized trial, with tissue-specific expression profiling (GTEx), was then used to explore the consistency of these computational predictions at the protein and tissue levels. Statistical analysis was performed using appropriate parametric and nonparametric tests, including ANCOVA where applicable.
    Results: DDG reduced HbA1c compared with placebo (-0.32%, P=0.032). Fasting plasma glucose showed a borderline reduction (P=0.050). Network pharmacology identified 617 potential targets intersecting with 2,652 DEGs, yielding 29 candidates. Using machine-learning combined with protein-protein interaction topology and literature support, we further prioritized eight core targets (P2RX7, IL1B, PTPN1, AKT2, CD38, NFE2L2, NOS3, and MERTK). Enrichment analyses of these candidates, together with serum proteomic profiling, implicated PI3K-Akt signaling, inflammatory and oxidative stress responses, and focal adhesion-related pathways.
    Conclusion: Clinically, DDG used as add-on therapy to metformin produced a modest but statistically significant improvement in glycemic control in patients with inadequately controlled T2DM. Our findings are consistent with the hypothesis that DDG may act through a multi-target network spanning inflammatory (P2RX7, IL1B), insulin/metabolic (PTPN1, AKT2, CD38), oxidative-endothelial (NFE2L2, NOS3) and vascular-resolution (MERTK) axes, generating testable mechanistic hypotheses for future experimental studies.
    Keywords:  machine learning; multi-omics integration; network pharmacology; proteomics; traditional chinese medicine; type 2 diabetes mellitus
    DOI:  https://doi.org/10.3389/fphar.2025.1723584
This page is being maintained by Thomas Krichel. It was last updated on 2026‒01‒25 at 0:13.