Citations

Objective We aimed to use machine learning (ML) models to investigate the impact of clinical, social and behavioural factors on 1-year progression from pre-diabetes to type 2 diabetes mellitus (DM). Design A retrospective cohort study. Setting A large health system including eight sites in southeast Michigan. Participants Adults with haemoglobin A1c (HbA1c) between 5.7% and 6.4% for two consecutive years between 1 January 2008 and 31 December 2023, and no prior history of type 2 DM or metformin use. Primary outcome measure New-onset type 2 DM (HbA1c ≥6.5%) in 1 year. Results Among 11 809 individuals, 815 (6.9%) progressed to type 2 DM within 1 year. CatBoost demonstrated the best performance (average area under the curve 0.78). Prior-year HbA1c was the most influential covariate (SHapley Additive exPlanations 1.02). Traditional metabolic factors (high-density lipoprotein, body mass index (BMI), white blood cell, age, gender, triglycerides) also contributed. Lastly, while inclusion of social and behavioural determinants of health (SBDH) did not significantly improve the overall model performance, depression emerged as the prominent SBDH covariate. Depression was a stronger predictor of diabetes progression in individuals with higher baseline BMI and lower baseline HbA1c. Conclusions Inclusion of social and behavioural covariates provided no incremental value for prediction of diabetes progression from pre-diabetes. However, machine learning revealed that depression may play a role in progression to type 2 DM.

Objectives To understand the trade-offs between different statistical modeling approaches, using real world data with small sub-populations, with rare exposures and increasingly rare outcomes. In particular, to compare adjusted regression, inverse probability weighting, and matching. Methods Data for these analyses came from the RADAR (N=1,134) and combined CNICS/JHHCC (N=14,434) cohorts. We estimated prevalence ratios (PRs) for self-reported use of specific substances comparing subpopulations (SP), SP-1 vs SP-3 and SP-2 vs SP-3, where SP-1 was the largest proportion (92% in RADAR, 81% in CNICS/JHHCC), SP2 was moderate proportion (18% in CNICS/JHHCC) and SP-3 was the smallest proportion (8% in RADAR, 1% in CNICS/JHHCC) of the population. We calculated PRs using 1) unadjusted relative risk regression (RR); and adjusted estimates controlling for age, race/ethnicity, study site, and year of interview using: 2) standard adjustment in RR; 3) stabilized inverse probability of treatment weighting (IPTW); and 4) matching with up to 3 matches from SP-1 or SP-2 per SP-3 participant. Results For most substances, all methods yielded consistent estimates. There were large weights in some of the IPTW analyses and in three cases these resulted in substantially divergent estimates. For the comparison between SP-1 and SP-3, the estimate for smoking was 1.3-fold greater in the matched analysis (PR=1.33, 95% CI: 1.02-1.75) than in IPTW (PR=1.03, 95% CI: 0.78-1.37 ATE and PR=1.05, 95% CI: 0.88-1.26 ATT). Even more extreme divergence in estimates was observed for differences between SP-2 and SP-3 with respect to methamphetamine/amphetamine, (IPTW-ATE: PR=1.51, 95% CI: 0.79-2.89; IPTW-ATT: PR=2.36, 95% CI: 1.36-4.12); vs Matching: PR=2.71, 95%CI: 1.47-4.99) and cocaine (IPTW-ATE: PR=1.15, 95% CI: 0.55-2.38; IPTW-ATT: PR=1.81, 95% CI: 1.03-3.18); vs Matching: PR=1.38, 95%CI: 0.76-2.53). Conclusion The combination of a rare exposure and a rare outcome can produce challenges for commonly used confounding adjustment strategies, and it is often best to compare different modeling approaches to gain greater insight.

Background: Aortic endograft infection (AEGI) is a rare complication of abdominal endovascular aortic repair (EVAR) and thoracic endovascular aortic repair (TEVAR). Surgical explantation with reconstruction is recommended, but data on medical management remain limited. We present our institutional experience with AEGI. Objective: To describe the clinical presentation, microbiology, management strategies, and outcomes of patients with AEGI, with the aim of addressing current gaps in the literature and informing future research to improve patient outcomes. Design: Retrospective descriptive study. Methods: Adult patients admitted with AEGI at our institution between January 2017 and August 2024 were identified using ICD-10 codes and confirmed using the Management of Aortic Graft Infection Collaboration (MAGIC) diagnostic criteria. Clinical, microbiological, radiographic, treatment, and outcome data were collected and analyzed. Results: We screened 517 patients, and 46 met AEGI criteria (EVAR 54%, TEVAR 46%). Most (63%) presented with late-onset infections (⩾3 months). Fever (54%) and sepsis (48%) were common presenting symptoms. Computed tomography (CT) (85%) was the primary imaging modality. Blood cultures were positive in 54% of the cases, with Staphylococcus aureus being the most common monomicrobial pathogen (24%), and 24% of cases were polymicrobial. The majority (67%) received combined medical and surgical therapy. Chronic antibiotic suppressive therapy was used in 34 (74%), most often tetracyclines (33%). Overall, treatment success rates were similar between surgical and medical groups (58% vs 60%, p = 0.250) with comparable mortality. However, relapses were more frequent in the medical-only group (33% vs 16%, p = 0.185). Conclusion: Most patients presented with late-onset AEGI. Fever was absent in nearly half of cases, and blood cultures were positive in only 54%. Overall, surgical and medical therapy had similar success, though relapses were more frequent in those treated medically alone.

We demonstrate the structural and functional similarities of organoid-derived gastric mucus and native mucus collected from human patients. The bioengineered gastric mucus mimics its native counterpart in its proteomic profile and physical architecture. This work highlights the translational potential of organoid-derived mucus for functional investigations of the human gastric mucus layer.

Abstract: Extracorporeal membrane oxygenation (ECMO) is increasingly used to support patients with severe cardiac or respiratory failure. Neurological complications occur across ECMO configurations and represent a major contributor to mortality and long-term disability. Major neurological complications include ischemic and hemorrhagic stroke, hypoxic ischemic brain injury, seizures, delirium, spinal cord and peripheral nervous system injury, and brain death. The mechanisms underlying neurological injury during ECMO are multifactorial and are influenced by ECMO configuration, reflecting differences in cerebral perfusion, gas exchange, anticoagulation exposure, and precannulation physiologic insults. This review synthesizes neurological complications associated with ECMO, neurological monitoring, and outcomes.

Understanding the complex interplay between electrical activity and neurochemical signaling is essential for advancing neuromodulation therapies, yet current technologies often measure these modalities in isolation and suffer from technical limitations such as stimulation artifacts. To address these challenges, we developed the Multifunctional Apparatus for Voltammetry, Electrophysiology, and Neuromodulation (MAVEN), a compact, battery-powered platform designed for high-resolution, concurrent multimodal data acquisition. Preclinical validation in both small- and large-animal models demonstrated that MAVEN provides stable, high-fidelity multimodal readouts while effectively minimizing stimulation artifacts during simultaneous programmable neurostimulation. By capturing phasic and tonic neurochemical changes, alongside real-time electrophysiological data, we established the system's feasibility for continuous monitoring. Beyond its utility in investigating the mechanisms of neurological and neuropsychiatric disorders, MAVEN has been engineered as a versatile analytical platform. Recent successful applications in monitoring pharmacological agents, such as vancomycin, further demonstrate its potential as a robust point-of-care device for real-time diagnostic and therapeutic monitoring. These results position MAVEN not only as a powerful tool for developing personalized, closed-loop neuromodulation strategies but also as a next-generation analytical device capable of broader clinical and bio-sensing applications.

Abstract Background Quality assurance (QA) in radiation therapy is a critical component of ensuring accurate and consistent patient treatments, but many tasks are time consuming and rely on human visual acuity, which limits their accuracy. Purpose We aimed to automate monthly mechanical QA tests, improve measurement accuracy and precision, and decrease inter‐user variability using a computer vision‐based quality assurance (CVQA) system. Methods A custom marker board incorporating four ArUco markers was created along with a custom camera holder that mounted onto the gantry head. OpenCV was utilized to automate tests for couch translation, collimator and table angles, collimator and table walkout, optical distance indicator (ODI), and field size detection. A GUI was created to guide users through the tests, determine passing status of measurements based on MPPG8.b criteria, and display image captures to allow users to troubleshoot if needed. Reproducibility tests were taken across four days. The system was tested against manual measurements by graph paper or digital level for each test. Field sizes and ODI were read by 12 different physicists to determine human variability and comparisons for CVQA. Results The CVQA system took 5 min to set up and 7 min to perform all tests. All field sizes, collimator/table angles, collimator/table walkout, and table translations were reproducible within 0.5 mm and 0.5°. ODI measurements were reproducible within 1 mm. Table travel, ODI, and walkout measurements agreed with manual measurements within 0.5 mm and 0.4°, except for vertical table motions that agreed within 0.9 mm due to the lens focus being optimized for the 100 cm SSD plane. The standard deviation between physicists for almost all symmetric and asymmetric jaws was larger than the reproducibility of CVQA. CVQA has been utilized for five years and has demonstrated ability to identify mechanical machine issues. Conclusions CVQA successfully automates mechanical QA tasks, providing an efficient and precise system. CVQA is open source and freely available for academic institutions. Its adoption can improve workflow efficiency and consistency in clinical environments.

Background/Objectives: The PRO-CTCAE Average Composite Score (ACS), calculated as the mean of PRO-CTCAE composite scores, summarizes overall symptom and side effect burden. This study evaluated the test–retest reliability, responsiveness to change, known-groups validity, and sensitivity to group differences in the ACS. Methods: Analyses used the original PRO-CTCAE validation dataset, including lung (n = 174), breast (n = 222), and head and neck (n = 139) cancer cohorts, and data from the COMET-2 randomized phase III trial comparing cabozantinib (n = 53) and mitoxantrone–prednisone (n = 54) in men with previously treated prostate cancer. Test–retest reliability was assessed using intraclass correlation coefficients (ICCs) from two-way mixed-effects models for agreement. Responsiveness was evaluated by relating ACS change scores (Visit 1 minus follow-up Visit 2) to patient-reported global ratings of change (GRC; worsened, unchanged, or improved) using standardized response means (SRMs) and Jonckheere–Terpstra trend tests. Known-groups validity was examined by comparing mean ACS values between ECOG performance status groups (0–1 vs. 2–4). Sensitivity to treatment group differences was assessed using model-based area-under-the-curve (AUC) comparisons of longitudinal ACS trajectories. Results: Test–retest reliability was acceptable, with ICCs among GRC-defined stable patients of 0.80 (95% CI: 0.70–0.87) in lung cancer, 0.84 (95% CI: 0.77–0.89) in breast cancer, and 0.77 (95% CI: 0.63–0.86) in head and neck cancer; ICCs based on assessments completed one day apart ranged from 0.88 to 0.90. The ACS demonstrated responsiveness, with SRMs of 0.30/0.15/−0.37 (lung), 0.29/0.14/−0.40 (breast), and −0.01/−0.20/−0.56 (head/neck) for improved/no-change/worsened groups, respectively, and significant monotonic trends across GRC categories (all p < 0.01). Known-groups validity was supported by conceptually expected differences in ACS values across distinct levels of self-reported patient-reported physical functioning and ECOG performance status categories. Mean ACS values were lower among patients with good versus limited physical functioning (0.71 vs. 1.23 in lung cancer, 0.52 vs. 1.19 in breast cancer, and 0.69 vs. 1.29 in head and neck cancer) and among patients with ECOG PS 0–1 versus 2–4 (0.93 vs. 1.31, 0.74 vs. 1.22, and 0.90 vs. 1.06, respectively). In COMET-2, higher symptom burden was detected in the cabozantinib arm compared with the mitoxantrone–prednisone arm (AUC difference = 1.5, 95% CI: 0.2–2.8, p = 0.02). Conclusions: The ACS demonstrated acceptable test–retest reliability, responsiveness, and known-groups validity across multiple cancer populations. These findings support its use as a complementary summary measure of overall symptomatic adverse event burden alongside individual symptom-level analyses.

Background and Objectives: We developed the Mayo Outpatient Precepting Evaluation Tool (MOPET) to assess and improve clinical teaching for family medicine residents. Given the scheduling challenges of peer evaluation, we sought to improve the MOPET’s usefulness by evaluating the reliability of self- and resident-reported teaching behaviors during clinical precepting compared to peer observation. Methods: Forty-eight half-day sessions in a family medicine residency clinic were assessed with parallel observer, preceptor, and resident MOPET forms. Interrater reliability was calculated using the κ (kappa) statistic. Results: Residents reported that preceptors used 14 of 20 teaching behaviors more often than observers. Observers and preceptors significantly agreed on the use of five teaching behaviors, while residents agreed with observers on two teaching behaviors. Conclusions: Although self-assessment and resident-based feedback of preceptor teaching behavior may improve practicality, they are less reliable compared to direct peer observation with the validated MOPET.

Importance Ischemic heart disease (IHD), the leading cause of death in the US, is predominantly due to modifiable risk factors. Estimates of IHD mortality attributable to risk factors provide evidence for health policy decision-making. Objective To estimate the burden of IHD death attributable to risk factors in the US from 1990-2023. Design, Setting, and Participants The Global Burden of Disease Study 2023 (GBD 2023) used vital records and a broad set of epidemiologic data to estimate IHD death rates, risk factor exposure, and relative risk curves for risk-outcome pairs for 1990-2023 for the general population. Data analysis was performed from October 2024 to December 2025. Exposure Twelve metabolic, behavioral, and environmental risk factors. Main Outcomes and Measures The primary outcomes were IHD death rates per 100 000 persons, counts, and attributable risks from 1990-2023 by age, sex, and US state. Estimates include 95% uncertainty intervals (UI). IHD death rates were estimated using ensemble modeling methods. Risk exposures were estimated using bayesian meta-regression methods. Relative risks were estimated following the Burden of Proof framework. Results In 2023, there were 473 000 IHD deaths (95% UI, 414 000-510 000) in the US, a decrease of 58.7% (95% UI, 56.8%-61.0%) in age-standardized rate since 1990. Between 2010 and 2023, there was a 19.0% decrease (95% UI, 15.0%-22.9%) for males and a 24.5% decrease (95% UI, 20.3%-29.7%) for females in IHD death rates. High systolic blood pressure (SBP), dietary risks, and high low-density lipoprotein cholesterol (LDL-C) were the leading risk factors for IHD deaths in 2023, accounting for 47.2% (95% UI, 36.4%-57.0%), 38.6% (95% UI, 17.2%-56.8%), and 28.5% (95% UI, 19.3%-39.6%) of IHD deaths, respectively. Increased exposure to several risk factors substantially increased their attributable burden for IHD deaths in 2023, with high fasting plasma glucose (FPG) increasing 38.8% (95% UI, 11.5%-81.1%) and high body mass index (BMI) increasing 54.5% (95% UI, 41.8%-66.3%) since 1990. Smoking and particulate matter pollution had the greatest decrease in attributable IHD mortality since 1990, at 33.3% (95% UI, 23.6%-41.7%) and 74.9% (95% UI, 46.7%-88.8%), respectively. Conclusion and Relevance Per the results of this systematic analysis of GBD 2023, a total of 88.7% (95% UI, 83.4%-92.5%) of IHD deaths were attributable to modifiable risk factors in the US in 2023, with high SBP, dietary risks, and high LDL-C being the greatest contributors. High BMI and high FPG showed the largest attribution increases, while exposure to other risks did not increase significantly for the population.