Nevin Manimala

Genome Biol. 2026 Apr 23. doi: 10.1186/s13059-026-04066-2. Online ahead of print.

ABSTRACT

High-throughput single-cell sequencing is widely used to study cell identity. We present SEAGALL (Single-cell Explainable Geometry-Aware Graph Attention Learning pipeLine), a deep learning method to quantify the impact of molecular features on cellular phenotype, based on geometry-regularised autoencoders (GRAE) and explainable graph attention networks (X-GAT). The GRAE embeds the data into a latent space to build a reliable cell-cell graph. The GAT is trained to learn the annotations and XAI is used to explain the predictions, unravelling the features driving cell identity. SEAGALL extracts specific and stable signatures from multiple omics experiments, going beyond differential marker genes.

PMID:42026624 | DOI:10.1186/s13059-026-04066-2

BMC Psychiatry. 2026 Apr 23. doi: 10.1186/s12888-026-07973-7. Online ahead of print.

NO ABSTRACT

PMID:42026587 | DOI:10.1186/s12888-026-07973-7

BMC Oral Health. 2026 Apr 23. doi: 10.1186/s12903-026-08440-1. Online ahead of print.

NO ABSTRACT

PMID:42026580 | DOI:10.1186/s12903-026-08440-1

BMC Health Serv Res. 2026 Apr 23. doi: 10.1186/s12913-026-14607-x. Online ahead of print.

ABSTRACT

BACKGROUND: Healthcare workers are exposed to sustained occupational stressors that may lead to heterogeneous patterns of psychological distress and resilience, with potential implications for job satisfaction. Person-centered approaches may help identify subgroups with distinct mental health profiles.

METHODS: We conducted a cross-sectional survey between December 2024 and March 2025 among healthcare workers employed in a single Italian Local Health Authority (ASL 1 Abruzzo). Participants completed measures of anxiety (GAD-7), depressive symptoms (PHQ-9), insomnia (ISI), resilience (RSA-11), perceived stress (PSS-10), and job satisfaction (JSS). K-means clustering on standardized mental health measures was used to identify profiles; the optimal number of clusters was selected using multiple internal validation indices (Gap statistic, average silhouette width, elbow method, and NbClust majority rule). Clusters were compared on sociodemographic variables and job satisfaction using non-parametric tests and chi-square tests.

RESULTS: Of 383 respondents, five were excluded (n = 2 for invalid response patterns, n = 3 for incomplete data), yielding N = 378. A two-cluster solution emerged. Cluster 1 (n = 145) showed higher psychological distress and perceived stress and lower resilience; Cluster 2 (n = 233) showed lower anxiety, depressive symptoms, insomnia and perceived stress, alongside higher resilience. Job satisfaction was significantly higher in Cluster 2 than Cluster 1 (Wilcoxon W = 21521, p < 0.001, r = 0.23). Cluster membership also differed by gender and work site.

CONCLUSIONS: Two distinct mental health profiles were identified within a single health authority, highlighting a subgroup characterized by high distress, low resilience and lower job satisfaction. Targeted psychosocial and organizational interventions may be warranted to support this vulnerable group and sustain workforce wellbeing.

PMID:42026564 | DOI:10.1186/s12913-026-14607-x

BMC Public Health. 2026 Apr 23. doi: 10.1186/s12889-026-27331-2. Online ahead of print.

NO ABSTRACT

PMID:42026562 | DOI:10.1186/s12889-026-27331-2

BMC Infect Dis. 2026 Apr 23. doi: 10.1186/s12879-026-13375-7. Online ahead of print.

NO ABSTRACT

PMID:42026521 | DOI:10.1186/s12879-026-13375-7

J Neurodev Disord. 2026 Apr 23. doi: 10.1186/s11689-026-09695-z. Online ahead of print.

NO ABSTRACT

PMID:42026485 | DOI:10.1186/s11689-026-09695-z

Radiography (Lond). 2026 Apr 22;32(4):103416. doi: 10.1016/j.radi.2026.103416. Online ahead of print.

ABSTRACT

INTRODUCTION: Missed fractures in the Emergency Department (ED) can lead to delayed treatment and patient harm. Radiographer preliminary clinical evaluation (PCE) aims to support referrers when interpreting radiographs in the absence of a definitive clinical report. This pre-implementation study evaluated radiographer participation and diagnostic accuracy in a department without an existing radiographer abnormality detection system.

METHODS: A prospective service evaluation study was conducted in a general hospital. Radiographers were asked to provide PCE for consecutive ED musculoskeletal trauma radiographs. Participation was recorded. PCEs were compared with the clinical report, and sensitivity, specificity and accuracy were calculated with 95% confidence intervals. Accuracy was evaluated against a fixed performance standard using a non-inferiority test. Differences in proportions between the first and second halves of the study period, in participation and accuracy, were assessed using z-tests.

RESULTS: Of 937 eligible examinations, 412 contained a PCE comment (44.0% participation), increasing significantly from 39.3% (182/463) in the first half of the study period to 48.5% (230/474) in the second half (p = 0.0045). After exclusions, 369 PCEs were analysed. Sensitivity, specificity and accuracy were 80.2%, 94.2% and 89.3%, respectively. The lower bound of the accuracy confidence interval (85.7%) exceeded the non-inferiority margin (82% accuracy), confirming PCE accuracy was statistically non-inferior to the 92% benchmark. Median time from examination attendance to PCE entry was 12 min.

CONCLUSION: Radiographers provided timely, accurate PCEs, achieving performance comparable with published standards, demonstrating the feasibility of implementing a PCE service.

IMPLICATIONS FOR PRACTICE: This study contributes to the evidence supporting PCE in departments without a 24-h hot-reporting service. Even without a dedicated training package, radiographers performed well, but this study emphasises that sensitivity remains a key area for further improvement.

PMID:42026441 | DOI:10.1016/j.radi.2026.103416

Nurse Educ Today. 2026 Apr 17;163:107122. doi: 10.1016/j.nedt.2026.107122. Online ahead of print.

ABSTRACT

BACKGROUND: Digital readiness has become a critical competency for future nurses in the evolving landscape of healthcare. Two essential components of this readiness-eHealth literacy and attitudes toward artificial intelligence (AI)-have gained prominence in nursing education. However, limited evidence exists regarding their interrelationship and associated demographic and behavioral factors, particularly in middle-income countries.

AIM: This study aimed to assess eHealth literacy and attitudes toward AI among undergraduate nursing students, identify associated demographic and behavioral factors, and examine the relationship between these two constructs as key dimensions of digital readiness.

METHODS: A descriptive and cross-sectional analytic study was conducted with 286 undergraduate nursing students at a public university in Türkiye during the 2024-2025 academic year. Participants were recruited using a voluntary, open invitation approach. Data were collected through an online survey, including a sociodemographic questionnaire, the eHealth Literacy Scale (eHEALS), and the General Attitudes Toward Artificial Intelligence Scale. Non-parametric tests and Spearman’s rho correlation were used for group comparisons and bivariate associations. Hierarchical multiple regression analysis was performed to evaluate the independent contribution of eHealth literacy to AI attitudes while controlling for potential confounding demographic and behavioral variables.

RESULTS: Of the 286 participants, the majority were female (75.9%) with a mean age of 20.7 years (SD = 2.78), predominantly in the 19-22 age range. Participants reported moderate-to-high eHealth literacy (M = 28.3, SD = 4.92) and favorable attitudes toward AI (M = 65.9, SD = 8.37). Significant differences were observed across gender, academic year, grade point average (GPA), and frequency of AI tool use. Students using generative AI tools such as ChatGPT scored significantly higher on both scales. A positive correlation was found between eHealth literacy and AI attitudes (r = 0.214, p < 0.001). In hierarchical regression analysis, demographic variables accounted for a small proportion of variance in AI attitudes, while behavioral factors (e.g., GPA and AI tool use) significantly improved the model. However, eHealth literacy did not make a statistically significant independent contribution to AI attitudes after controlling for these variables.

CONCLUSION: Undergraduate nursing students demonstrated promising levels of digital readiness. However, the relationship between eHealth literacy and AI attitudes appears to be context-dependent rather than independently predictive. Behavioral engagement with digital technologies plays a more prominent role in shaping AI attitudes. These findings underscore the need for nursing curricula to move beyond foundational digital literacy and incorporate experiential, practice-oriented AI learning opportunities to support comprehensive digital readiness.

PMID:42026438 | DOI:10.1016/j.nedt.2026.107122

Behav Res Methods. 2026 Apr 23;58(5):137. doi: 10.3758/s13428-026-02999-x.

ABSTRACT

In nonexperimental studies, obtaining an unbiased estimate of the average treatment effect (ATE) typically requires two key assumptions: that all relevant covariates are measured (i.e., no unmeasured confounding) and that the statistical model used for covariate adjustment is correctly specified. Two common approaches for adjustment are specifying an outcome model and propensity score weighting. To mitigate bias from model misspecification, doubly robust methods combine both approaches, ensuring unbiased ATE estimates if either the outcome model or the propensity score model is correctly specified. In this study, we review four doubly robust methods that have received considerable attention in the methodological literature but remain underutilized in psychological research: augmented inverse probability weighting, regression weighted by the inverse propensity score, regression incorporating the inverse propensity score as a covariate, and calibrated propensity score weighting. Using two simulation studies, we compare these methods with regression estimation and inverse probability weighting estimators. Our results suggest that doubly robust methods-particularly regression weighted by the inverse propensity score-offer greater protection against bias from model misspecification across various data-generating scenarios. We also discuss practical considerations for implementing doubly robust methods, including weight normalization, propensity score truncation, and potential efficiency losses due to overfitting. The different methods for estimating the ATE are illustrated in a data example.

PMID:42026423 | DOI:10.3758/s13428-026-02999-x