To systematically evaluate the quality and readability of health information generated by four large language models (LLMs) in response to inquiries regarding type 2 diabetes mellitus (T2DM), using an authoritative Chinese clinical guideline as the reference standard.

A total of 124 standardized questions were extracted from the Chinese Type 2 Diabetes Popular Science Guidelines. Six endocrinologists and diabetes specialists conducted independent, blind evaluations using the CLEAR tool (Completeness, Lack of false Information, Evidence, Appropriateness, Relevance) and PEMAT-P (Patient Education Materials Assessment Tool for Printable materials). Response characteristics were also recorded. Between-model differences were tested using the Kruskal-Wallis H test with Bonferroni pairwise comparisons.

All four models achieved total CLEAR scores within the "very good" range (19-25), with no significant differences seen between models (χ² = 1.985, p = 0.576). No significant differences were observed in the dimensions of Lack of false information (χ² = 7.644, p = 0.054), Evidence (χ² = 2.309, p = 0.511), and Relevance (χ² = 7.516, p = 0.057). However, significant differences emerged in Completeness (χ² = 47.661, p < 0.001) and Appropriateness (χ² = 88.360, p < 0.001). Claude-4.0 received the lowest score in Completeness (median 4.00, IQR 3.00-5.00) but achieved the highest ranking in Appropriateness (median 4.00, IQR 4.00-5.00). On the PEMAT-P, understandability differed significantly across models (χ² = 159.120, p < 0.001), yet all models surpassed the 70% threshold, with ChatGPT-4.1 highest (median 91.91%, IQR 91.91-100.00%). However, despite significant differences among the various models (χ² = 354.023, p < 0.001), only ERNIE Bot 4.5 Turbo (median 75.00%, IQR75.00-75.00%) surpassed the 70% threshold, with no single model demonstrating consistent superiority across all dimensions.

Although the four LLMs generally provide accurate and pertinent information regarding type 2 diabetes, enduring limits in actionability and inconsistencies among models in content completeness and understandability restrict their effective use in diabetic patient education. Future development should prioritize stronger step-by-step behavioral guidance and differentiated, scenario-specific model deployment to enhance their value in patient-facing diabetes self-management support.

To compare the efficacy and safety of glucagon-like peptide 1 (GLP-1) based drug treatments for weight loss in adults with overweight or obesity without diabetes.

Network meta-analysis of randomised controlled trials.

Embase, PubMed (Medline), and Web of Science, 1 January 2000 to 6 March 2026.

Randomised controlled trials that enrolled adults with overweight or obesity, comparing GLP-1 receptor agonists or related co-agonists with placebo or active comparators, with a minimum intervention duration of 12 weeks. Excluded were trials that enrolled participants with diabetes, or where diabetes status could not be clearly determined.

58 trials of 24 214 participants were analysed. Compared with placebo, weight loss was greatest with retatrutide (-22.10%, 95% confidence interval -25.60% to -18.60%), followed by tirzepatide (-19.28%, -20.39% to -18.16%), and CagriSema (a combination of cagrilintide and semaglutide, -17.32%, -19.32% to -15.32%). Conventional GLP-1 receptor agonists showed more modest effects. Similar patterns were seen for waist circumference and lipid outcomes. Treatment rankings suggested a probabilistic hierarchy favouring next generation incretin based treatments, although confidence intervals overlapped for several comparisons. Low certainty evidence suggested higher rates for discontinuing treatment with danuglipron and retatrutide, whereas mazdutide showed better tolerability.

In adults with overweight or obesity without diabetes, next generation incretin based treatments achieved greater weight loss than conventional GLP-1 receptor agonists. Differences in tolerability, limited head-to-head evidence, and residual uncertainty, however, should be considered when interpreting comparative treatment effects.

PROSPERO CRD420261279841.

About one-quarter of patients with acute pancreatitis (AP) will develop diabetes mellitus (DM) within 3 years, but risk factors remain unclear. This study aims to determine whether machine learning models (ML) can be trained to accurately predict new-onset DM following AP and identify key clinical features using real-world data.

This retrospective cohort study used de-identified data from the TriNetX federated electronic health records (EHR) network from 1 January 2017 to 11 March 2024. A total of 58 746 patients with AP (International Classification of Diseases-10 code K85) and no prior diagnosis of DM were included. New-onset DM following AP was the main outcome of interest. Five ML models were trained across four prediction windows, including logistic regression (LR), eXtreme Gradient Boosting, Adaptive Boosting, Random Forest and support vector machine. Model performance was assessed using the area under the receiver operating characteristic curve (AUROC).

Among the 58 746 patients with AP (mean (SD) age, 50.2 (16.6) years), the LR model demonstrated the highest overall performance, with a mean accuracy of 0.72 (SD, 0.009) and an AUROC of 0.79 (SD, 0.009). Key clinical features across models included age, pancreatic necrosis, body weight, body mass index, systolic blood pressure, number of medical visits and prior AP laboratory values such as glucose, anion gap, blood urea nitrogen (BUN) and total protein.

In this first real-world evidence study using EHR data, we have developed and demonstrated the feasibility of using ML to predict the new onset of DM after AP. Clinical features such as age, pancreatic necrosis, prior glucose, BUN and anion gap had the highest overall importance scores and may inform tailored prevention strategies.

Lipohypertrophy (LH) is a common complication of insulin-treated diabetes; however, its prevalence and risk factors in people using continuous subcutaneous insulin infusion (CSII) remain insufficiently characterized. This study aimed to determine the prevalence and severity of LH and to identify the associated risk factors in individuals with type 1 diabetes mellitus (T1DM) treated with CSII or multiple daily injections (MDI).

A total of 121 individuals with T1DM who received insulin therapy for at least one year were included (CSII, n = 43; MDI, n = 78). The presence and severity of LH (non-LH, mild, moderate, or severe) were assessed using the methods described by Ucieklak et al. Clinical characteristics were extracted from medical records, and diabetes self-management variables were collected using self-reported questionnaires.

The distribution of LH severity did not differ significantly between the CSII and MDI groups (CSII: 41.9% non-LH, 20.9% moderate, and 37.2% severe; MDI: 46.2%, 17.9%, and 35.9%, respectively; p = 0.908). In the CSII group, diabetic peripheral neuropathy (DPN) was associated with higher odds of LH (OR 12.3, 95% CI: 1.3-1651; p = 0.025). In addition, the total daily insulin dose differed significantly across LH severity categories (p = 0.049). In the MDI group, individuals with LH had a younger age at diagnosis and tended to have a longer duration of diabetes and insulin treatment than those without LH.

LH is highly prevalent among individuals with T1DM treated with either CSII or MDI. Distinct risk factors were identified according to the insulin delivery method, with DPN associated with severe LH in CSII users, and longer disease duration associated with LH in MDI users. These findings support targeted screening strategies to improve LH detection and prevent related complications.

The online version contains supplementary material available at https://doi.org/10.1007/s13340-026-00928-z.

Hashimoto's thyroiditis (HT) and diabetes mellitus are highly prevalent chronic immune-mediated disorders that frequently co-occur and share genetic susceptibility, T-helper (Th) 1/Th17 skewing, and regulatory T-cell (Treg) dysfunction. Among individuals with type 1 diabetes mellitus (T1DM), autoimmune thyroiditis is the most common comorbid autoimmune disease. Extracellular vesicles (EVs) have emerged as important mediators linking autoimmune and metabolic inflammation. This review compares how EVs remodel the immune microenvironment in HT, type 2 diabetes mellitus (T2DM), and related disease contexts, with attention to donor cells, cargo, recipient pathways, biomarkers, and therapeutic implications. In HT and T1DM, EVs can deliver organ-specific autoantigens, whereas in classical T2DM current evidence more strongly supports EVs as carriers of stress signals, chemokines, and immunoregulatory miRNAs that shape islet inflammation and insulin resistance rather than autoantigen presentation; latent autoimmune diabetes in adults is considered separately. Across these diseases, recurrent EV-miRNA programs and DAMP/NLRP3 signaling converge on Treg/Th17 imbalance and M1/M2 macrophage polarization. We also emphasize the marked asymmetry of evidence maturity, with substantially stronger in-vivo and clinical support on the T2DM side than on the HT side. This asymmetry is treated as an explicit interpretive boundary throughout the review. We assess circulating and urinary EV cargoes as liquid-biopsy candidates and discuss EV-based drug delivery, engineered immunomodulatory EVs, and modulation of EV biogenesis. Translational claims remain limited by heterogeneity, manufacturing, and safety challenges, particularly in organ-specific autoimmune disease. In this review, the term "immune-metabolic crossroads" refers to shared mechanisms, shared biomarker opportunities, and partially overlapping therapeutic entry points.

Impaired wound healing remains a major complication following calcaneal fracture surgery in patients with type 2 diabetes mellitus (T2DM), yet reliable tools for individualized risk prediction are lacking.

Of 553 T2DM patients screened, 360 were eligible and randomly allocated to a training set (n = 269 after excluding 1 for incomplete outcome adjudication; 66 outcome events) and an internal validation set (n = 90; 22 outcome events, 24.4%). A nomogram was developed using multivariable logistic regression based on predictors selected via LASSO. Model performance was assessed by area under the ROC curve (AUC), calibration (evaluated by the Brier score and the Hosmer-Lemeshow goodness-of-fit test), and decision curve analysis (DCA).

The final model incorporated five predictors: HbA1c, insulin use, diabetic peripheral neuropathy, C-reactive protein, and platelet count. In the training set, the nomogram achieved an apparent AUC of 0.886 (95% CI: 0.846-0.925),with an optimism-corrected AUC of 0.875 (optimism estimate = 0.011). In the internal validation set, the AUC was 0.891 (95% CI: 0.789-0.965). The model showed favorable calibration in the training set [Brier score = 0.113 (95% CI: 0.088-0.140); Hosmer-Lemeshow test, χ ² = 7.215, df = 8, P = 0.514] and good calibration in the validation set (Brier score = 0.106 (95% CI: 0.067-0.147);calibration slope = 1.27 (95% CI: 0.75-2.89),intercept = 0.44 (95% CI:-0.24-1.85); Hosmer-Lemeshow test, χ ² = 10.552, df = 8, P = 0.228), and provided higher net benefit than treat-all or treat-none strategies across threshold probabilities from 0.01 to 0.95in both the training and validation sets. At a threshold of 0.4, it identified approximately one-fifth of patients as high-risk, with a high proportion of events among them.

The internally validated nomogram provides a promising tool for risk stratification of postoperative wound complications in T2DM patients with calcaneal fractures. The model is exploratory and limited to internal validation; external validation in multicenter cohorts is required before routine clinical adoption.

To develop and validate machine learning models and an individualized nomogram for predicting pedicle screw loosening after posterior lumbar interbody fusion (PLIF) in osteoporotic patients using preoperative clinical, imaging, and bone metabolism-related medication profiles.

A retrospective analysis was conducted on 630 osteoporotic patients who underwent PLIF at three spine surgery centers. Patients were divided into a non-loosening group (n = 450) and a loosening group (n = 180) according to the presence of implant loosening on imaging within 12 months postoperatively. Univariate analysis was used to screen candidate variables, and LASSO-logistic regression with 10-fold cross-validation was applied to extract independent predictors. Four models (naïve Bayes, logistic regression, linear discriminant analysis, and decision tree) were constructed based on the selected features and evaluated using the area under the curve (AUC), calibration curves, and decision curve analysis (DCA). The structure of the optimal model (decision tree) was visualized, and a nomogram was built using multivariable logistic regression.

Univariate analysis showed significant differences between the two groups in age, bone mineral density T-score, pelvic incidence, lumbar lordosis, sex, hypertension, diabetes mellitus, foraminal morphology, history of glucocorticoid use, calcium supplementation, and vitamin D supplementation (all P < 0.05). LASSO regression identified 11 independent predictors (λ.min = 0.0325). Among the four models, the decision tree showed the highest discrimination, achieving an AUC of 0.975 (95% CI 0.961-0.989) in the training set and 0.971 (95% CI 0.955-0.987) in the test set, with good calibration (Hosmer-Lemeshow test P = 0.418). DCA demonstrated a significant net benefit across a clinically relevant threshold probability range (0-50%). The decision tree identified the bone mineral density T-score as the root splitting variable; a T-score < -3.6 classified patients as being at extremely high risk for loosening. Among those with T-score ≥ -3.6 who did not take calcium, lack of vitamin D supplementation was associated with a very high risk. Among those with T-score ≥ -3.6 who took calcium, female sex combined with lumbar lordosis ≥ 51° indicated an elevated risk. The nomogram integrated the 11 factors and yielded a C-index of 0.928 (95% CI 0.903-0.953) with good calibration (Hosmer-Lemeshow test P = 0.372).

The decision tree model demonstrated favorable predictive performance for pedicle screw loosening after PLIF. Its interpretable classification rules facilitate rapid screening of high-risk patients, while the nomogram enables individualized probability estimation for surgical planning. Together, these complementary tools offer a potentially useful framework for preoperative risk stratification and personalized management of osteoporotic patients undergoing PLIF.

Insulin resistance is important in the cause of type 2 diabetes mellitus (T2DM), but traditional markers for obesity are not enough to explain metabolic risk. The idea for this review stems from an interest in understanding the interplay between ectopic fat accumulation and activity of skeletal muscle with gravitational loading in precipitating insulin resistance and a 'Gravity-Muscle-Ectopic Fat Axis' as an integrative framework. We conducted a literature search using a narrative review approach in PubMed, Scopus, Web of Science, and Google Scholar. We identified and screened published studies from 2018 to 2026 and synthesized up to 40 high-quality, representative articles. We critically analyzed the evidence on ectopic fat, Skeletal Muscle Metabolism, anti-gravity muscle activity, Exercise interventions, and glucose homeostasis. Ectopic deposition of fat in skeletal muscle, pancreas, and liver is thought to be associated with insulin resistance and impaired pancreatic β-cell function. Ectopic fat and insulin sensitivity are tightly linked, and metabolic health is closely tied to both. Skeletal muscle has emerged as a key regulator of glucose homeostasis, and emerging evidence suggests that gravitational loading and anti-gravity muscle activation may contribute to more favorable ectopic fat distribution and improved metabolic regulation. This Gravity-Muscle-Ectopic Fat Axis offers a new conceptual model that connects muscle activity, ectopic fat loss, and insulin sensitivity. Addressing muscle health and ectopic fat may be promising strategies for preventing and treating insulin resistance (IR) and T2DM.

Low-dose CT lung cancer screening reduces lung cancer mortality in individuals at high risk and frequently reveals clinically relevant incidental findings. In the National Lung Screening Trial, approximately one-third of screening CT examinations demonstrated abnormalities other than lung cancer, raising challenges related to interpretation, reporting, and further management. The absence of consistent guidance and the variable clinical significance of these findings may hinder the broader adoption of lung cancer screening. Opportunistic screening leverages low-dose CT examinations performed for lung cancer detection to identify additional serious conditions at no extra cost, radiation exposure, or patient burden. The authors provide a practical framework for recognizing, reporting, and managing common opportunistic findings on lung cancer screening CT images, with an emphasis on evidence-based recommendations and current professional society guidance. Key thoracic findings include emphysema, interstitial lung abnormalities, interstitial lung disease, coronary artery calcification, thoracic aortic aneurysm, and pulmonary artery enlargement. Each of these findings, if unrecognized, is associated with substantial morbidity and mortality. Extrapulmonary findings such as low bone density and lesions found incidentally in the upper abdominal organs further expand the clinical value of lung cancer screening. The authors highlight standardized visual and quantitative assessment methods, recommended follow-up pathways, and the importance of structured reporting, including appropriate use of the Lung CT Screening Reporting and Data System "S" (other non-lung cancer findings) modifier. Emerging artificial intelligence tools are discussed as critical enablers for scalable opportunistic screening, offering automated detection, quantification, and longitudinal assessment of emphysema, coronary calcium levels, vascular dimensions, bone density, body composition, and selected abdominal pathologic conditions. By integrating opportunistic screening into routine lung cancer screening workflows, radiologists can extend the benefit of lung cancer screening with CT beyond cancer detection to facilitate comprehensive risk stratification and preventive care.

Primary sinonasal mucosal melanoma (SNMM) is a rare and highly aggressive malignancy that is frequently diagnosed at advanced stages because of its nonspecific presentation. We report the case of a 70-year-old Peruvian woman who presented with a six-month history of recurrent unilateral epistaxis refractory to direct compression, chemical cauterization, and nasal packing, associated with progressive nasal obstruction, facial pain, foul-smelling nasal discharge, and significant weight loss. Nasofibrolaryngoscopy revealed a friable pigmented mass arising from the left nasal cavity. Computed tomography demonstrated a lesion originating from the inferior turbinate without evidence of adjacent bone invasion. Histopathological examination showed an epithelioid malignant neoplasm with moderate pigmentation and a high mitotic index (11 mitoses/mm²), and immunohistochemistry was positive for SOX10, confirming the diagnosis of primary SNMM. This case highlights the diagnostic challenge of SNMM and emphasizes that persistent unilateral epistaxis refractory to conventional treatment should prompt early endoscopic evaluation to exclude structural lesions, including uncommon sinonasal malignancies.