Category: Statistics

BMC Infect Dis. 2026 May 25. doi: 10.1186/s12879-026-13648-1. Online ahead of print.

ABSTRACT

BACKGROUND: Blood transfusion is an essential component of healthcare delivery but poses a risk of transfusion-transmissible infections (TTIs), particularly in low- and middle-income countries where donor screening systems may be inadequate. This study aimed to determine the prevalence of major TTIs and assess their association with sociodemographic characteristics among blood donors at a tertiary care hospital in Lahore, Pakistan.

METHODS: A retrospective cross-sectional study was conducted using blood donor records of 550 individuals. All donations were routinely screened for hepatitis B virus (HBV), hepatitis C virus (HCV), human immunodeficiency virus (HIV), syphilis, and malaria. Prevalence rates were calculated using descriptive statistics. Associations between TTIs and donor characteristics were evaluated using chi-square tests. Univariate and multivariate logistic regression analyses were performed to identify independent predictors of infection.

RESULTS: The overall prevalence of TTIs was 12.18%. HBV was the most prevalent infection (4.73%), followed by HCV (4.18%), syphilis (2.55%), and HIV (0.73%). No cases of malaria were detected. Significant associations were observed between selected TTIs and donor characteristics, particularly age and donor status. Multivariate logistic regression analysis confirmed that these sociodemographic factors remained independently associated with infection status.

CONCLUSIONS: A substantial burden of transfusion-transmissible infections persists among blood donors in this setting. Strengthening donor selection, improving screening strategies, and promoting voluntary non-remunerated blood donation are essential to enhance transfusion safety and reduce the risk of TTIs in resource-limited settings.

PMID:42185824 | DOI:10.1186/s12879-026-13648-1

BMC Med Inform Decis Mak. 2026 May 25. doi: 10.1186/s12911-026-03584-0. Online ahead of print.

ABSTRACT

BACKGROUND: A significant challenge in blood donation is the occurrence of adverse donor reactions (ADRs) and their subsequent negative impact on the blood supply and public health. A promising strategy to mitigate these events is the deployment of non-invasive, cost-effective artificial intelligence (AI) models for donor screening and monitoring. This study aims to systematically review the AI models utilized in the identification and prediction of ADRs.

METHODS: This study used a systematic review approach in line with the PRISMA 2020 guidelines. A search was performed across various databases, including Web of Science, PubMed, Embase, Google Scholar, and Scopus. The results were combined narratively and presented using descriptive statistics. The quality of eligible studies was assessed using the Newcastle-Ottawa Scale (NOS) tool.

RESULTS: Among the 13 studies, nine were classified as immediate reactions and four as delayed reactions. The commonly used models included regression models, classical statistical models, and machine learning algorithms such as Random Forests, Gradient Boosting Machines (GBMs), XGBoost, and Artificial Neural Networks (ANNs). The main standard evaluation metrics for the models included Odds Ratio, Precision-Recall Area Under the Curve (PR-AUC), F1 Score, Precision, and Recall.

CONCLUSIONS: Adverse reactions among blood donors negatively impact donor retention and, by extension, the stability of the blood supply for patient care. In this context, AI models may offer a promising tool for supporting the prediction and monitoring of ADRs. However, the available studies are not sufficient to support the widespread adoption of these models in clinical or operational decision-making. Heterogeneity in study design, outcomes, and evaluation metrics together with limitations such as limited implementation, risk of bias, unclear reference standards, and a lack of external validation has constrained the interpretability and generalizability of the findings. Therefore, future research with more rigorous study designs, standardized reporting, harmonized evaluations, and external validation is essential to establish the effectiveness and reliability of these models.

PMID:42185820 | DOI:10.1186/s12911-026-03584-0

BMC Cancer. 2026 May 25. doi: 10.1186/s12885-026-16154-4. Online ahead of print.

ABSTRACT

BACKGROUND: Artificial intelligence (AI) is increasingly recognized as a valuable tool for the early detection and prognosis of oral cancer, addressing the challenge of high mortality due to late diagnosis. Artificial intelligence based diagnostic models have the potential to improve accuracy in differentiating between malignant, premalignant and benign oral lesions. This systematic review and meta-analysis evaluated the diagnostic performance of non-imaging and image-based artificial intelligence models and narratively synthesized evidence on prognostic and risk stratification applications in oral cancer.

METHODS: This study follows PRISMA guidelines to ensure quality and reproducibility. A systematic search across PubMed, Embase, Web of Science, Google Scholar and Scopus identified studies from 2010 to 2024 on artificial intelligence applications in oral cancer diagnosis. Sixteen eligible studies met predefined inclusion criteria, including AI-based screening compared to histology. Data extraction and bias assessment were conducted independently using QUADAS-2. The findings highlight AI’s potential in early detection and prognosis, emphasizing the need for further validation and clinical integration to enhance diagnostic accuracy.

RESULTS: A total of 801 studies were initially identified, with 53 undergoing further review, ultimately selecting 16 studies. Sample sizes varied from 70 to 44,000, allowing a broad evaluation of AI’s diagnostic performance. Artificial intelligence models showed wide range of sensitivity (42%-100%), specificity (63%-100%), and accuracy (63%-100%). Meta-analysis revealed a pooled sensitivity of 0.90 (95% CI: 0.81-0.98), specificity of 0.89 (95% CI: 0.84-0.95), and accuracy of 0.89 (95% CI: 0.83-0.95), with substantial heterogeneity (I² = 100%). Image-based models had higher pooled sensitivity (0.94 vs. 0.76, P = 0.320), specificity (0.93 vs. 0.79, P = 0.025), and accuracy (0.93 vs. 0.81, P = 0.042).

CONCLUSIONS: Artificial intelligence models show promising diagnostic performance for oral cancer based on retrospective clinical data. Although image-based models, particularly convolutional neural networks, demonstrated higher pooled sensitivity and specificity than non-imaging models, these differences were not statistically significant. Results should be interpreted with caution due to substantial heterogeneity. Advances reported in the literature, such as multimodal approaches and data augmentation, may improve non-imaging model performance and help narrow the gap between methodologies. These developments highlight AI’s potential in enhancing early detection and prognosis of oral cancer.

PMID:42185818 | DOI:10.1186/s12885-026-16154-4

BMC Med Educ. 2026 May 25. doi: 10.1186/s12909-026-09503-1. Online ahead of print.

ABSTRACT

BACKGROUND: Palliative Care focuses on improving quality of life and preventing suffering, especially near the end of life. With an aging population that frequently seeks care in the Emergency Department (ED), it is crucial for Emergency Medicine (EM) physicians to be well versed in the skills needed for palliative care. However, many EM residency programs lack formal training in this area. This study aimed to develop, implement, and assess a year-long, simulation-based palliative care curriculum.

METHODS: This IRB-exempt observational cohort study was conducted from July 2023 to June 2024 at a large urban EM residency program. A group of content experts in palliative care and simulation-based education designed the curriculum utilizing the Kern model of curriculum development, integrating simulation, small group discussion, and didactic lectures. The curriculum was delivered episodically throughout the academic year during regular residency didactics. The curriculum’s impact on knowledge, attitudes, and practice patterns regarding palliative care was assessed using pre- and post- intervention surveys.

RESULTS: Fourteen EM residents completed both pre- and post-intervention surveys. There was a statistically significant improvement in self-reported practice patterns (35.7% to 60.7% p = 0.0055), but no significant change in attitudes (73.4% to 73.4% p = 0.5020) or knowledge (48.9% to 57.1% p = 0.0758).

CONCLUSIONS: A year-long simulation-based palliative care curriculum significantly improved EM residents’ practice patterns in regard to palliative care, with a positive trend in knowledge retention. Feasibility of implementation was demonstrated. Findings suggest that simulation-based training effectively teaches key palliative care skills. This curriculum may serve as a model for integrating palliative care education into other EM residency programs and may improve resident preparedness in caring for critically ill patients near the end of life.

PMID:42185816 | DOI:10.1186/s12909-026-09503-1

BMC Nephrol. 2026 May 25. doi: 10.1186/s12882-026-05066-x. Online ahead of print.

ABSTRACT

BACKGROUND: Patients with end-stage kidney disease (ESKD) and acute kidney injury on chronic kidney disease (AKI-on-CKD) experience significant symptom burden and high mortality, yet evidence for structured palliative integration within nephrology remains limited in low- and middle-income countries.

AIM: To describe the timing and reasons for referrals from Nephrology to Palliative Medicine, the interventions delivered, and clinical outcomes following referral.

METHODS: A retrospective cohort study was conducted of all adult patients (n = 325) referred from Nephrology to Palliative Medicine. Data included demographics, renal diagnosis, dialysis modality, comorbidities, treatment intent, and documented palliative interventions, and were analyzed using descriptive statistics.

RESULTS: Mean age was 63.6 ± 14.1 years; 229 (70.5%) were male. Renal diagnoses comprised ESKD (215; 66.2%), AKI-on-CKD (86; 26.5%), and AKI (16; 4.9%). Dialysis modalities included maintenance hemodialysis (194, 59.7%), dialysis for AKI (63, 19.4%), and conservative kidney management (21, 6.5%). Referrals were primarily for GOC discussions (41.5%), followed by pain and symptom management (29.2%) and end-of-life (EOL) care (27.4%). Referrals specifically for conservative kidney management were uncommon (1.8%). Among deaths, 154 (75%) occurred in the hospital. Common triggers for referral included sepsis, multiorgan failure, and decisions to stop dialysis.

CONCLUSION: Palliative care involvement predominantly occurred during acute decline-often related to sepsis, multiorgan failure, or dialysis withdrawal. Although most patients died in the hospital, the high frequency of GOC and EOL discussions highlights the need for earlier, proactive palliative care engagement to support informed decision-making, symptom relief, and care aligned with patient preferences.

PMID:42185815 | DOI:10.1186/s12882-026-05066-x

BMC Nephrol. 2026 May 25. doi: 10.1186/s12882-025-04723-x. Online ahead of print.

ABSTRACT

BACKGROUND: CKD-related anemia remains a major complication in patients receiving peritoneal dialysis (PD), with limited treatment options beyond erythropoiesis-stimulating agents. Roxadustat, an oral hypoxia-inducible factor prolyl hydroxylase inhibitor, has shown promise in correcting CKD-related anemia and modulating iron and metabolic parameters. However, its evidence in PD remains limited.

METHODS: We conducted a systematic search of PubMed, Scopus, and Web of Science from inception to July 20, 2025. We included studies reporting the efficacy and safety of roxadustat in PD. Statistical analysis was performed using Review Manager (RevMan 5.4 for Windows) and R Studio.

RESULTS: Eight studies were included in the meta-analysis (n = 607). Roxadustat significantly increased hemoglobin at all time points (MD 0.35 g/dL, 95% CI 0.28-0.41; p < 0.00001), serum iron (MD 0.95 µmol/L, 95% CI 0.02-1.89; p = 0.05), and total iron-binding capacity (MD 6.25 µmol/L, 95% CI 3.95-8.55; p < 0.00001), and reduced hepcidin (MD – 12.28 ng/mL, 95% CI – 21.06 to – 3.50; p = 0.006). No significant effects were observed for ferritin (p = 0.49), transferrin saturation (p = 0.45), cholesterol (p = 0.07), LDL (p = 0.14), HDL (p = 0.27), triglycerides (p = 0.26), CRP (p = 0.75), systolic blood pressure (p = 0.10), or diastolic blood pressure (p = 0.08). Heterogeneity was low to moderate for most outcomes.

CONCLUSION: Roxadustat is effective in improving hemoglobin and iron metabolism in patients with PD, while exerting neutral effects on lipids, inflammation, and blood pressure. These findings support its role as a promising therapy for CKD-related anemia in PD.

CLINICAL TRIAL NUMBER: Not applicable.

PMID:42185813 | DOI:10.1186/s12882-025-04723-x

Horm Behav. 2026 May 25;182:105950. doi: 10.1016/j.yhbeh.2026.105950. Online ahead of print.

ABSTRACT

Some theories have proposed that children with autism spectrum disorder (ASD) may be exposed to increased levels of androgens during prenatal development, resulting in greater androgenization of testosterone-dependent traits including both peripheral (bodily) and central (brain-related) traits. Empirical support for this hypothesis is scant and inconsistent. In the present work, we studied an ostensible anatomical marker of testosterone exposure (sexually differentiated finger lengths) which develops during the first trimester or early second trimester of gestation. Participants were 25 boys with classic autism, recruited from the clinical practice of a local physician specializing in childhood autism, who met the standardized DSM (Diagnostic and Statistical Manual) criteria for ASD, and 57 normally-developing age- and sex-matched male and female controls (32 males, 25 females); N = 82; M_age = 7.30 yrs., SD = 4.18. Finger length was measured to the nearest 0.5 mm from digital images of the ventral surface of the hands using a validated measurement protocol. Consistent with past reports from adult samples, several of the finger length ratios were confirmed to display sex differences among control children, but the group of boys with ASD showed a female-like finger growth pattern, not the hypermasculine pattern predicted, and were found to differ statistically from the male controls. Boys with ASD thus showed a demasculinized pattern of finger differentiation. Our data do not support theories which suggest that greater fetal testosterone exposure occurs in boys with autism.

PMID:42184479 | DOI:10.1016/j.yhbeh.2026.105950

Neural Netw. 2026 May 19;203:109151. doi: 10.1016/j.neunet.2026.109151. Online ahead of print.

ABSTRACT

Accurate identification of the seizure onset zone (SOZ) is critical for the diagnosis and treatment of drug-resistant epilepsy (DRE). In recent years, although phase-amplitude coupling (PAC) has played an important role in epilepsy-related studies, few investigations have focused on applying PAC methods to SOZ identification. To this end, leveraging the capability of PAC to characterize neural interactions within the brain, this study computes the modulation index (MI) from clinical electrocorticography (ECoG) recordings of DRE patients. Subsequently, a statistical analysis of temporally evolving distributions of MI values across multiple frequency bands is conducted to analyze the differences in MI distribution features between SOZ and non-seizure onset zone (NSOZ) regions. Finally, distribution features of MI values are integrated with machine learning techniques to systematically evaluate the influence of different frequency bands and time windows on SOZ identification performance. The results demonstrate that MI distribution features can achieve accurate SOZ identification, with classification accuracy reaching 90.69%, indicating their potential as biomarkers for SOZ identification.

PMID:42184467 | DOI:10.1016/j.neunet.2026.109151

Neural Netw. 2026 May 19;203:109153. doi: 10.1016/j.neunet.2026.109153. Online ahead of print.

ABSTRACT

In safety-critical domains ranging from medical diagnostics to credit scoring, learning algorithms for set-structured data must satisfy strict theoretical axioms: permutation invariance, monotonicity, and computational scalability. However, deep learning architectures and axiomatic aggregation theory remain poorly connected. Building on recent bounds for sum-decomposable set representations, we formalize and specialize a fundamental efficiency limitation in canonical additive architectures (Deep Sets): for exact representation of basic order-statistics with an injective sum-decomposable encoder, sum-pooling requires linear feature scaling (m=Ω(N)), rendering them dimension-inefficient for winner-take-all logic. Conversely, standard attention mechanisms violate monotonicity via competitive normalization. To bridge this gap, we introduce the Monotone Set Transformer (MoST), a verifiable neuro-fuzzy architecture. MoST utilizes a dual-prong design: (1) a learnable non-competitive gating mechanism based on positive kernels to approximate supermodular synergies, and (2) a semantic anchor based on Ordered Weighted Averaging (OWA) that enables explicit rank-dependent aggregation with O(Nlog N) complexity. We prove that MoST preserves set-inclusion monotonicity by construction. In controlled synthetic settings aligned with the theory, MoST can realize near-exact max aggregation, while supplementary experiments show strong monotonic certification and favorable time/memory scaling versus attention baselines. Across molecular tasks, results highlight a task-dependent trade-off between strict monotonic constraints and unconstrained predictive flexibility.

PMID:42184462 | DOI:10.1016/j.neunet.2026.109153

Microbiol Res. 2026 May 22;310:128560. doi: 10.1016/j.micres.2026.128560. Online ahead of print.

ABSTRACT

Heavy metal (HM) contamination in mining soils threatens ecological security and agricultural sustainability worldwide. Phytoremediation is frequently constrained by low plant accumulation efficiency and poor environmental adaptability of plant growth-promoting bacteria in multi-metal contaminated environments. The microbial mechanisms underlying Paenibacillus polymyxa WZ14 enhances multi-HMs (Cd, Pb, Cu) accumulation in diverse leguminous plants remain unclear. A pot experiment combined with 16S rRNA high-throughput Illumina MiSeq sequencing and multivariate statistical analyses was coupled with investigate the regulatory effects of WZ14 inoculation on strengthening leguminous plant (Robinia pseudoacacia L., Sophora xanthantha, Cajanus cajan L., and Albizia kalkora Roxb) HMs accumulation and the underlying microbial mechanisms. In this study, WZ14 inoculation significantly increased the concentrations and total uptake of Cd, Pb and Cu in four leguminous plants, with the strongest enhancement effect on S. xanthantha (Cd, Pb, and Cu uptake increased by 92.78%, 66.26%, and 117.92%, respectively). This promotion effect exhibited distinct plant species specificity and HM type dependence. Besides, WZ14 reshaped rhizosphere microbial community structure, increased relative abundance of HM-responsive dominant genera (Sphingomonas and Flavobacterium increased by 5.06%-31.03% and 8.64%-73.05%, respectively), enhanced microbial co-occurrence network modularity and cooperative interactions. PLS-PM analysis further clarified that available phosphorus, Sphingomonas and Flavisolibacter were the key factors regulating plant HMs uptake, and confirmed the chain regulatory pathway of soil nutrients-dominant genera-plant HMs uptake. Conclusively, this study clarifies the microbial-mediated regulatory mechanism of HMs uptake between P. polymyxa WZ14 and leguminous plants, providing a reliable and environmentally friendly new strategy for the ecological remediation of multi-HM contaminated soils.

PMID:42184460 | DOI:10.1016/j.micres.2026.128560