Workplace violence against healthcare workers is a serious threat to safety, efficiency, and health system performance worldwide. This study looks at the causes and effects of workplace violence in Pakistan’s hospital sector, where public and private facilities face very different challenges. We surveyed 768 healthcare workers from six hospitals in Peshawar—three public and three private. We collected detailed information on their experiences with violence, the quality of their institutions, how they report incidents, and the personal effects over a twelve-month period. Our findings show a clear divide between public and private hospitals. Workers in public hospitals face physical violence more than four times as often as those in private hospitals. More than half of public hospital staff reported experiencing attacks, compared to just 12 percent in private facilities. This divide appears in all areas: Witnessing violence, how incidents are reported, and the availability of safety measures and formal complaint systems. We created a model to explain how poor institutional quality leads to high levels of violence and low reporting, and we confirmed these predictions with our data. Statistical analysis shows that factors like hospital type, security measures, and reporting procedures account for almost all variations in violence outcomes. Individual characteristics such as gender, profession, and experience do not have an independent impact. Having formal reporting systems increases the actual reporting of incidents by nearly ten times. We also found that workers who are physically attacked face significant psychological issues, including heightened fear of work and dissatisfaction with their jobs. This suggests serious challenges for keeping workers in their roles. These findings indicate that structural problems, rather than individual weaknesses, are the main cause of workplace violence in healthcare settings in Pakistan. Policymakers should focus on creating mandatory reporting systems, investing in visible security measures, and offering psychological support to affected staff, especially in the public hospital sector, which is severely under-resourced.

Background: Fetal growth restriction (FGR) is associated with haemodynamic redistribution that preferentially reduces renal perfusion. Renal artery Doppler pulsatility index (PI) has been proposed as a non-invasive marker of this redistribution; however, clinically applicable diagnostic thresholds have not been established in Indian populations. Methods: This prospective comparative study enrolled 40 singleton pregnancies between 28 and 38 weeks of gestation: 20 with confirmed FGR (Group A) and 20 uncomplicated controls (Group B). Doppler PI and resistance index (RI) of bilateral fetal renal arteries, the middle cerebral artery, and the umbilical artery were measured using a 3.75-MHz curvilinear transducer. Receiver operating characteristic (ROC) curve analysis was performed to determine the optimal diagnostic PI threshold using Youden’s index. Results: Renal artery PI was significantly elevated in FGR for both the right kidney (2.57 ± 0.18 vs. 1.56 ± 0.16; p < 0.001) and left kidney (2.49 ± 0.18 vs. 1.59 ± 0.22; p < 0.001), with a large effect size (Cohen’s d ≥4.4). ROC analysis yielded an area under the curve (AUC) of 1.000 (95% CI: 0.999–1.000) for the right renal artery and 0.999 (95% CI: 0.990–1.000) for the left. A PI threshold of >2.10 for the right renal artery and >2.00 for the left offered sensitivity and specificity exceeding 97% for the diagnosis of FGR in this cohort. Renal artery RI did not differ significantly between groups. Fetal kidney dimensions correlated strongly with gestational age in both groups with no intergroup size difference. Conclusion: Fetal renal artery PI demonstrates excellent discriminatory performance for FGR, with near-perfect AUC values and diagnostically actionable thresholds (right renal PI >2.10; left renal PI >2.00). These thresholds warrant prospective validation in larger multicentre cohorts before integration into routine clinical surveillance protocols.

Background: Young adults, including medical students, are increasingly exposed to unhealthy lifestyle behaviors and early metabolic abnormalities that may predispose them to long-term cardiovascular disease. This constitutes a major public health challenge in South East Asia now-a days. Methods: A cross-sectional study was conducted among undergraduate medical students at Noakhali Medical College, Bangladesh, between October 2024 and June 2025. Behavioral, anthropometric, clinical, and biochemical data were collected using standardized protocols. Lifestyle factors included diet, physical activity, sedentary behavior, sleep duration, and stress. Anthropometric measurements comprised body mass index (BMI) and waist circumference, while biochemical assessments included 75-gm oral glucose tolerance test (OGTT), glycated hemoglobin (HbA1c) and lipid profile. Descriptive statistics and bivariate analyses using Chi-square tests were performed. Results: A total of 135 students (age: 21.0±1.9 years; 69.6% female) were included. The prevalence of overweight/obesity (BMI ≥25 kg/m²) was 28.9%, and similar numbers had central obesity. More than half of participants exhibited prediabetes (51.9%), while 64.4% had dyslipidemia and 13.3% met criteria for metabolic syndrome. Hypertension was present in 4.4% of students. Overweight/obesity was significantly associated with hypertension (p = 0.002) and dyslipidemia (p = 0.033), but not with sex, physical inactivity, sedentary behavior, dietary intake, glucose intolerance, or short sleep duration. Conclusion: Undergraduate medical students in Bangladesh demonstrate a high burden of cardiometabolic risk factors, including excess body weight, early dysglycemia, and dyslipidemia. These findings underscore the need for early, targeted preventive and health promotion strategies within medical education to mitigate future cardiometabolic disease risk in this critical population.

Background: Colorectal cancer (CRC) is the second commonest cause of cancer-related death in the United Kingdom and the third commonest cancer worldwide. Preoperative chemoradiotherapy is widely used to down-stage rectal tumours and improve surgical outcomes; however, its long-term molecular consequences on normal rectal tissue remain poorly understood. Objective: To identify molecular changes in gene and protein expression in normal rectal tissue of CRC patients who received preoperative chemoradiotherapy compared to those who underwent surgery alone. Methods: Normal rectal muscle-layer tissue samples were collected from 14 patients (chemoradiotherapy group, n=7; surgery-only control group, n=7). RNA was extracted and used to synthesise cDNA for RT-qPCR analysis of 15 target genes involved in autophagy, oxidative stress, neuro-axonal transport, angiogenesis, and tissue elasticity. Protein expression of GPX3 was assessed by western blotting. Statistical analysis was performed using the Mann-Whitney U test. Results: Variable changes in gene expression were observed between the two groups, most notably in ATG7, CAT, PRCKD, and GPX4. None of these differences reached statistical significance (p > 0.05), although CAT showed a trend (p=0.073). GPX3 gene expression demonstrated a statistically significant difference (p=0.024), but western blotting revealed no corresponding significant change in GPX3 protein expression (p=0.80). Conclusion: Preoperative chemoradiotherapy appears to alter gene expression in the oxidative stress pathway in normal rectal tissue, though these changes were not statistically significant due to the small sample size. Larger studies with quantitative protein analysis are warranted to confirm these findings.

The readmission of patients into hospitals has been considered a measure of the quality of care delivered, the quality of discharge planning, and continuity of care. In Saudi Arabia, different digital health initiatives and advancements in electronic health record systems offer a timely opportunity to synthesize evidence on different artificial intelligence-based models for predicting hospital readmission risk and identify key design features of the model that are clinically relevant and useful. In this article, we have synthesized evidence on different artificial intelligence-based models for predicting hospital readmission risk and healthcare utilization outcomes using different machine learning and deep learning techniques and different explainable artificial intelligence techniques for predicting short-term hospital readmission risk and healthcare utilization outcomes within a 7-30-day window using peer-reviewed articles published between 2020 and 2025. The evidence on different artificial intelligence-based models for predicting hospital readmission risk and healthcare utilization outcomes using different machine learning and deep learning techniques and different explainable artificial intelligence techniques is discussed in detail with special emphasis on its applicability to hospital systems in Saudi Arabia. The gradient boosting and tree-based ensemble techniques are found to be consistently performing well for structured electronic health record-based predictive models. The representation learning-based techniques are found to be performing well for structured electronic health record-based predictive models. The outcome definition heterogeneity, class imbalance, temporal leakage, and lack of external validation are some of the limitations identified in different studies on artificial intelligence-based models for predicting hospital readmission risk and healthcare utilization outcomes. For hospital systems in Saudi Arabia, different factors need to be considered for developing artificial intelligence-based models for predicting hospital readmission risk and healthcare utilization outcomes. The methodological blueprint for hospital systems in Saudi Arabia is proposed using different artificial intelligence-based models for predicting hospital readmission risk and healthcare utilization outcomes based on different machine learning and deep learning techniques and different explainable artificial intelligence techniques using the PRISMA 2020 protocol and TRIPOD+AI protocol for systematic reviews and reporting of prediction model performance.

Nigeria accounts for 28.3% of global maternal deaths, with a maternal mortality ratio of 1,047 per 100,000 live births, the highest nationally estimated figure globally. This study assessed reproductive health (RH) service preparedness across 143 primary health care (PHC) facilities in Edo State, Nigeria, constructed a validated nine-item RH Composite Preparedness Score (CPS), and identified structural determinants of preparedness deficits using multivariate regression. A census-based cross-sectional facility survey was conducted across all 18 Local Government Areas of Edo State (23–25 September 2020) using the SOML Integrated Supportive Supervision checklist, aligned with the WHO Service Availability and Readiness Assessment framework. Ordinary Least Squares and binary logistic regression were applied alongside chi-square and one-way ANOVA tests to examine geographic variation and structural predictors. The mean RH CPS was 4.92/9 (SD = 1.31), indicating that facilities lacked over half of essential RH items on average. Partograph availability was critically deficient at 17.5% and maternal death audit forms at 14.7%, representing gaps of 82.5% and 85.3% respectively. No significant inter-district variation was detected, indicating systemic rather than geographically localized failure. OLS regression was non-significant (R² = 0.031; p = .743), demonstrating that general infrastructure variables do not explain RH preparedness deficits. Among 14 secondary hospitals, partographs and maternal death audit forms were each absent in 57.1% of facilities. These findings indicate that RH preparedness failures are programmatically rather than infrastructure-determined, and are addressable through targeted procurement, competency-based training, and mandatory accountability mechanisms integrated into existing supervision systems.

This study presents a robust machine learning framework to enhance stroke risk prediction and support global health objectives to reduce stroke-related morbidity and mortality. Using the Kaggle Stroke Prediction Dataset, the research integrates advanced data preprocessing, exploratory data analysis, and five machine learning algorithms: Logistic Regression, Random Forest, Gradient Boosting, XGBoost, and LightGBM. A stacked ensemble model serves as the core methodology, achieving superior predictive performance (accuracy 97%, F1-score 0.97, AUC-ROC 0.99) compared with individual models and prior works. Bayesian Optimization ensures optimal hyperparameter selection, while explainability methods such as SHAP and LIME bolster model interpretability to meet clinical transparency demands. The methodology aligns with the NZ Ngā Tikanga Paihere Framework, embedding principles of data privacy, cultural sensitivity, and fairness. Addressing challenges such as data imbalance and computational scalability via SMOTE and distributed computing, the model demonstrates robust performance, validated through bootstrapping and cross-validation. This research advances ethical, accurate, and actionable AI-driven stroke prediction for healthcare applications.