Browse Articles
Discover research articles across all indexed journals
Design of an integrated model using U-Net, DeepSurv, and cross-attention for lung cancer classification and survival prediction
Abstract Lung cancer ranks within the highest mortality rates among cancerous diseases; hence, its detailed classification and survival rate prediction are of utmost importance. Most existing approaches for the classification and prognosis prediction in lung cancer share a critical deficiency: they are either single-modality or fail to learn complex, nonlinear interactions between distinct data types. However, none of these traditional models iteratively refines segmentation with the requisite accuracy to embed continuous flow of new patient data without degradation in performance. We hence propose an Iterative Multi-Model Deep Learning Framework for improved classification of lung cancer subtypes and predictions of survival rates. Our proposed work uses the U-Net model, which refines features extracted iteratively to improve precision in segmented regions. For lung cancer subtypes classification, feature-level fusion is done by using CNN for spatial features extracted from both radiological and histopathology images and using an MLP for genomic data samples. The DeepSurv model extends the Cox proportional hazards model with deep learning to handle complex, multi-dimensional clinical, imaging, and genomic data for survival rate prediction. Bayesian optimization is used to optimize the hyperparameter tuning process, whereas EWC empowers this approach with real-time survival predictions, thus enabling incremental learning without catastrophic forgetting. This is further reinforced by a multimodal attention mechanism that ensures the most discriminative features from each modality are taken into consideration by the model. The contributions of this work consist of an improvement in tumor segmentation accuracy with results that range from 90 to 95% Dice similarity, a raise of accuracy in lung cancer subtype classification between 85% and 90%, and robust survival rate predictions with a C Index of ~ 0.75–0.80. Besides, our adaptive learning approach can continuously improve our model to make it fit for real-time clinical applications. The framework will present an end-to-end solution for the diagnosis and prognosis of lung cancer.
Whole genome sequencing of a hypermucoviscous, multidrug-resistant Klebsiella pneumoniae subsp. pneumoniae K219 isolated from human sputum
A complex circular intuitionistic fuzzy decision framework for evaluating sustainable agriculture strategies under uncertainty
Distribution and antibiotic resistance patterns of airborne staphylococci in urban environments of Delhi, India
Noninvasive subterahertz glucose monitoring using a communication inspired eye diagram
Predicting surgical outcomes in spring assisted cranioplasty via finite element analysis and animal experiments
Abstract Sagittal craniosynostosis, a congenital cranial suture disorder, is treated with spring-assisted cranioplasty (SAC), but optimal surgical parameters remain unclear. This study explores computational models to predict surgical outcomes of SAC by linking biomechanics with bone regeneration and cranial remodeling. Fifteen 3-week-old Sprague–Dawley (SD) rats underwent SAC with nickel-titanium springs (0–100 g forces). Bone regeneration was tracked via fluorescence labeling, while micro-CT scans measured cephalic index (CI), bone mineral density (BMD), and bone volume fraction (BV/TV). Finite element analysis (FEA) simulated stress and strain distributions. Regression models were established to predict the relationship between mechanical indicators and surgical outcomes. The 50 g group achieved optimal bone regeneration and cranial correction, while > 80 g forces risked bone damage. FEA indicates stress and strain over the bone are influenced by spring force, rat geometry, and bone density. Regression analysis revealed linear strain-CI relationships and quadratic strain-BV/TV and strain-BMD relationship. Moderate spring force (50 g) or strain (1.0%) enhances osteogenesis without structural compromise. Computational models provide a biomechanically grounded framework for SAC optimization, advancing precision treatment for sagittal craniosynostosis. Future studies should validate findings in disease-specific models and assess long-term outcomes.
Comparative performance of ChatGPT-4o, ChatGPT-5, and gemini 2.5 flash on Persian internal medicine subspecialty board exams
Game theoretic modeling and optimization of competition and collaboration in dual channel electronic waste supply chains
The association between vegetable-derived nitrate and nitrite intake, cardiovascular risk factors and glycemic markers in obese individuals
Telmisartan reduces systemic inflammation and alters the renin-angiotensin system in mild COVID-19
Deployment and validation of predictive 6-dimensional beam diagnostics through generative reconstruction with standard accelerator elements
Multiscale and re-entrant surface analysis of multi jet fusion surfaces and the effect of postprocessing via sandblasting
Abstract The growing use of Multi Jet Fusion (MJF) 3D printing in industry brings challenges related to surface quality and post-processing effects. This study presents a comprehensive analysis of surface geometry changes in MJF-printed parts before and after sandblasting. A suitable abrasive material was selected and tested, followed by chemical composition analysis to confirm the absence of abrasive residues on the surfaces. Glass beads with a radius of 37.5 to 70 μm were used for sandblasting. Micro-computed tomography (µCT) was used as the only technique capable of capturing complex surface topography, including re-entrant features, both pre- and post-processing. Surface characterization was performed following ISO 25,178 roughness parameters and area-scale analysis for a total of 30 surfaces. A proprietary algorithm was also implemented to quantify re-entrant features. Results showed significant changes in surface texture due to sandblasting, including reduced surface roughness and enhanced uniformity. For instance, Sa decreased from 22.3 μm to 14.6 μm, and Sp dropped by over 30%. Multiscale analysis revealed a 50% reduction in surface fractal complexity ( Asfc ) and a significant decrease in fine-scale geometric variability. An algorithm for detecting reentrant features was developed, in which original indicators for their quantitative occurrence were introduced: r₁ and r₂. On average r₂ values dropped from 60% to 13%, and r₁ from 260% to 135%. ANOVA confirmed statistically significant differences ( p < 0.001) between pre- and post-processing states (MSR coefficients for Srel and Asfc parameters were used). Additionally, a statistical analysis was performed, which showed significant and large impact of post-processing on the geometric character of the surface. The study demonstrates that µCT is uniquely capable of isolating unsintered powder and characterizing complex surface features. Sandblasting significantly alters surface geometry and must be accounted for during design, especially in applications requiring precise topographical control.
Opportunities challenges and roadmap for humanoid robots in construction
The use of brain-specific biomarkers in urine for prediction of neurological outcome and extent of tissue damage following stroke
Abstract Serum biomarkers of neuronal and glial damage gained significant attention in recent years for their diagnostic and prognostic value in acute stroke. In this study, we explored the potential of spot urine testing as a practical, non-invasive alternative. Using ultrasensitive single-molecule arrays, we assessed levels of glial fibrillary acidic protein (GFAP), neurofilament light chain (NfL), ubiquitin carboxy-terminal hydrolase L1 (UCH-L1), and total tau (t-tau) in urine and blood samples from hospitalized patients with acute ischemic stroke (AIS) and intracerebral hemorrhage (ICH) within 96 h after symptom onset. The results were correlated with mortality rates, tissue damage extent and modified Rankin Scale (mRS) scores at discharge and after three months. Significant variables were included into an multivariate regression model with confounder correction (NIHSS at admission, age at onset and serum creatinine). As a result, absolute concentrations of urine GFAP, NfL and t-tau were significantly higher in patients with in-hospital death or death at 3 months follow-up compared to patients reaching a mRS ≤ 3. Using multivariate regressions models, urine GFAP demonstrated the strongest predictive value across all biomarkers for in-hospital functional outcome, which was superior to the respective GFAP serum concentrations, whereas NfL in serum was the only significantly predictive biomarker for 3 months functional outcome. For radiological outcomes, urine GFAP concentrations were closely linked with ischemic lesion size (ρ = 0.42, p = 0.01). In conclusion, the measurement of neuronal and glial biomarkers in urine is feasible and may enhance the prediction of functional and imaging outcomes.
Trends in childhood routine immunisation and maternal health service access and utilisation in Sokoto State, Nigeria. A quantitative survey
Abstract Sokoto State, Nigeria, continues to report some of the lowest maternal, neonatal, and child health (MNCH) indicators nationally. The COVID-19 pandemic further disrupted service delivery in Nigeria, yet its long-term impact on routine immunisation (RI) and maternal health services at the sub-national level in fragile and resource-constrained settings remains underexplored. We conducted a retrospective cohort study using secondary administrative data from 2015 to 2023 to examine the trends in the uptake of the first and third doses of the diphtheria-tetanus-pertussis containing vaccines (DTP-1 and DTP-3), and facility-based deliveries. Dynamic time-series regression models were applied to assess the influence of the COVID-19 period and contextual factors, including climate change, settlement type, and health budgets. DTP-1 coverage and proportion of facility deliveries improved over time. There was a significant increase in facility deliveries during the COVID-19 pandemic ( p = 0.019), while DTP-1 coverage declined modestly during the pandemic ( p = 0.030). DTP-3 showed no significant change ( p = 0.078). Urban residence was positively associated with facility delivery. Lagged dependent variables were strongly significant across the models, indicating path dependency in service utilisation. These findings highlight both resilience and vulnerability within Sokoto’s healthcare system. While maternal services were adapted during the pandemic, immunisation continuity remained fragile. These insights could inform the development of a pandemic-resilient framework for MNCH services in similar settings.
High concentrations of L-lysine cause mitochondrial damage and necrosis in isolated pancreatic acinar cells
A General Strategy to Access All Stereosequences in a Synthetic Polymer
Influence of substructure material on the scanning accuracy and scannability of implant-supported full arch bar substructures
Abstract Bar substructures may be utilized for fixed full-arch implant-supported restorations, where suprastructures are cemented as prosthetic shells. Certain clinical situations may necessitate intraoral scanning of the bar substructures. However, the scannability and the accuracy of the intraoral scans remain unclear. This study aimed to compare and assess the scannability and the scan accuracy of two different bar materials used for implant-supported prostheses. Two maxillary implant-supported substructures (bars) were milled from 2 different materials. Group I from titanium and Group II from poly-ether-ether-ketone (PEEK). The substructures were digitized by using a desktop scanner (Medit MD-1D0410, Medit Corp). The STL file produced was considered the reference. Each substructure was scanned intraorally ( n = 10) by using an intraoral scanner (IOS) (Medit i700; Medit Corp). The non-captured surface area in a preset time limit of 15 s (scannability) was assessed in (mm 2 ). To evaluate the scanning accuracy, all STL files were imported into a surface-matching software program (Medit Design v3.0.6 Build 286; Medit Corp), where overall RMS (root mean square) deviations in (µm) were calculated. The Shapiro–Wilk test of normality was used. A comparison between the study groups was done by using an independent samples t-test. Significance was set at P < .05. The titanium substructure demonstrated better trueness and precision concerning deviation analysis (202.40 ± 27.57 and 197.50 ± 24.69, respectively) than the PEEK substructure (262.20 ± 30.87 and 244.1 ± 9.18, respectively) ( P < .001). Regarding scannability, significantly more surface area was captured when scanning the PEEK substructure (985.42 ± 7.22) than titanium (951.15 ± 12.16) ( P < .001). PEEK was more scannable. The titanium bar demonstrated significantly higher accuracy than the PEEK bar, which was more scannable than the titanium bar. This trial, number NCT06423482, is registered at Clinical.gov. Its start date was May 14, 2024.