Clinician perspectives on the clinical utility of Belay Summit™ 2.0 cerebrospinal fluid test - A mixed methods study.

Harmonization in magnetic resonance imaging: A survey of acquisition, image-level, and feature-level methods.

Enhancement of OCT signal interpretability in deep penetration laser welding of aluminum

With the growing availability of bedside point-of-care technologies, an expanding body of evidence questions the necessity of routine daily chest radiographs in pediatric intensive care unit (PICU) patients. Nevertheless, this review illustrates that daily chest imaging continues to yield clinically meaningful information and should not be entirely discontinued. Routine daily chest radiographs in the PICU frequently reveal malpositioned devices including central venous catheters, chest tubes, nasogastric tubes, and endotracheal tubes. They may also uncover unanticipated cardiopulmonary abnormalities that warrant changes in management. Moreover, the potential risks of radiation exposure can be mitigated through the use of portable imaging and targeted scanning techniques that limit exposure to essential regions. Routine daily chest radiographs continue to be a valuable tool for pediatric intensivists and have the potential to identify unexpected complications that can significantly influence patient management. Future research should focus on risk stratifying pediatric patient populations which may benefit from routine chest radiography.

• Introduces a flow-aware training strategy for PINNs based on the information flow in the domain. • Decomposes the domain into geodesic subdomains that are progressively supervised via an epoch-based schedule. • Matches or improves baseline accuracy while reducing computational costs across seven PDE benchmarks. • Prevents uninformative updates and misleading gradients in regions where the solution is not yet determined. Physics-Informed Neural Networks (PINNs) typically update parameters at all collocation points from the first epoch, even in regions still unreachable from boundary or initial conditions. In these zones, early updates may be uninformative or harmful, as residual gradients often correlate poorly with the true error. We introduce a flow-aware training strategy that delays supervision until the governing physics can propagate meaningful information. The computational mesh is decomposed into geodesic subdomains ranked by distance from an information boundary, where initial or boundary conditions are applied, and progressively activated according to an epoch schedule. This selective exposure concentrates optimization where updates are most effective, preventing wasted capacity and misleading gradients in causally disconnected regions. The method requires no architectural changes and uses the standard PINN loss; only the sampling mask evolves over time. Benchmarks on seven PDE problems show that flow-aware training matches or improves baseline accuracy while reducing computational cost.

Advances in brain imaging have generated unprecedented volumes of high-dimensional data, yet extracting meaningful information from complex, noisy, and incomplete brain imaging data remains a significant challenge. Diffusion models (DMs) have introduced a paradigm shift in this field, surpassing traditional generative approaches. This review systematically examines the theoretical foundations of diffusion models, and their practical applications in eight brain imaging computing tasks: registration, super-resolution, cross-modal reconstruction and synthesis, segmentation, classification, brain network analysis, brain-computer interface (BCI) signals augmentation, and BCI decoding. Additionally, we emphasize obstacles that hinder deployment in practice, including computational scalability and sampling inefficiency, limited generalization under domain shift sensitivity, as well as multimodal integration and alignment constraints, while outlining potential future directions that emphasize the convergence of diffusion models with large-scale foundation models, which hold the potential to advance scalable, reliable, and clinically embedded brain imaging solutions. Throughout this review, we aim to establish a roadmap of progress and translational hurdles to guide emerging research and accelerate collaboration spanning DMs, clinical brain imaging, and engineering disciplines.

Harmonization in laboratory medicine is increasingly recognized as a fundamental requirement to ensure interchangeable laboratory information, avoid confusion among users, and ultimately improve patient safety. However, current evidence continues to highlight substantial gaps in harmonization across several dimensions of the total testing process, including test requesting strategies, analytical comparability among different methods and platforms, reporting units, reference intervals, and the clinical interpretation of results. The complete blood count (CBC) represents a paradigmatic example of the persistent lack of harmonization in laboratory medicine and of the potential clinical implications that may arise from it. The CBC is among the most widely requested and clinically impactful laboratory tests, being routinely used across nearly all areas of medical practice. However, recent papers published in highly influential medical journals, including Nature and JAMA, have highlighted two complementary but equally important issues. On the one hand, these studies emphasize the need for improved interpretative criteria, particularly through the adoption of personalized reference intervals that better reflect individual biological variability. On the other hand, they underline the substantial heterogeneity in the number and type of CBC parameters reported, which may generate uncertainty and confusion among clinicians regarding the clinical meaning and utility of the reported variables. Taken together, these observations highlight that harmonization should not be limited to analytical standardization alone but must also address the appropriateness of test requesting and the clinical interpretation of laboratory results. In this perspective, CBC may represent an ideal model for rethinking the role of laboratory medicine: moving from the simple generation of numerical results toward the delivery of actionable and clinically meaningful information. Achieving this goal requires stronger collaboration between laboratory professionals and clinicians, with the shared objective of aligning test requesting, reporting strategies, and interpretative criteria to the clinical question and the patient's context. Such an approach would represent a concrete step toward the implementation of Value-Based Laboratory Medicine, where the true value of laboratory testing is measured not by the volume of tests performed but by its ability to improve diagnostic accuracy, support clinical decision-making, and ultimately enhance patient outcomes.

Current robotic prostheses developed for individuals with transradial amputation often lack physiological feedback, particularly proprioceptive information, which limits control precision and increases reliance on vision. This study investigates the effect of introducing non-invasive somatotopic feedback using transcutaneous electrical nerve stimulation (TENS) to convey hand aperture information, as an artificial form of proprioceptive feedback. Twenty healthy participants with intact limbs were divided into two groups: one receiving TENS-based feedback and one without feedback. Participants performed an aperture control task under visual and non-visual conditions, with some trials including a concurrent Stroop task to assess cognitive load. We hypothesized that providing non-invasive somatotopic proprioceptive feedback via TENS mitigates key control and integration challenges, leading to improved accuracy, faster learning, and reduced reliance on vision, without increasing cognitive demands. Under visual deprivation, participants receiving TENS feedback achieved significantly smaller aperture control errors than those without feedback, both under alternating visual conditions (p < 0.001) and under prolonged visual deprivation conditions (p = 0.017). From the very first trial, TENS-based feedback enabled control accuracy comparable to that with visual input. Although perceptual shifts affected control accuracy under dual-task conditions, the +TENS group maintained high cognitive performance and effective control toward perceived functional targets, suggesting that the artificial feedback nonetheless supported hand control in the presence of a secondary task. These findings highlight the potential of noninvasive somatotopic proprioceptive feedback delivered through TENS to provide physiologically meaningful information and support future strategies for restoring proprioceptive function in prosthetic hand users.

Visualization of mutations and modifications on nucleotide and protein sequences provides insights into sequence hotspots conveying biologically relevant findings derived through genomics/proteomics platforms. Lollipop plots are suitable for such visualization, where structural modules can also be easily defined to further refine meaningful information in mutation/modification patterns. Although multiple tools are available for this purpose, they are limited by the data category that can be input and lack automation. We developed a multi-purpose, opensource, completely automated web-based tool that can visualize all possible omics data inputs feasible for generating ready-to-use lollipop plots. M2Viz is a web-based application designed for visualizing modifications and mutations (M2) identified from genomics/proteomics datasets on gene and protein sequences. It is a one-stop-solution for creating lollipop plots representing DNA methylation, Single Nucleotide Variations (SNVs), protein Post-Translational Modifications (PTMs), and Single Amino Acid Variations (SAAVs). Additionally, tool can be accessed to represent profiling and differential expression/regulation datasets, with options for visualizing numerical data. The tool provides end-to-end solution for uploading, processing, annotating and graphical representation. The system integrates a Python-Django backend for data handling, an R-based engine for advanced plotting, and a React.js frontend for a responsive and interactive user experience. Nucleotide or amino acid sequences and protein domain information are retrieved from Ensembl or UniProt through REST API, removing dependency on manual data downloading. M2Viz is instrumental in visualizing DNA methylation, mutational hotspots and PTMs on gene/protein sequences. It is a valuable tool across omics pipelines and bioinformatics. The visualization pipeline is automated to reduce the need for hosting sequence and structural information in the backend, along with custom plotting flexibility. M2Viz is freely accessible at "https://ciods.in/m2viz".

The aim of this study was to investigate the integrity of the optic nerve using the ratio between T1WI and T2WI (T1/T2) MRI signal intensities, and to assess whether inter-eye differences could indicate the presence of lesions in the optic nerve. The cohort included eighteen healthy controls and forty-nine individuals with multiple sclerosis; from those, thirty-two had a previous episode of optic neuritis. A method was trained to generate automatic segmentations of the optic nerve. Provided masks were then applied to the T1/T2 map and the mean profile was extracted. Inter-eye asymmetry was calculated from the area under the profiles. The presence of optic nerve lesions was confirmed using double inversion recovery MRI. Group differences were evaluated with non-parametric statistical tests, and the ability of inter-eye asymmetry to identify lesions was assessed using logistic regression. Automated masks showed a moderate agreement with the reference segmentations (mean Dice coefficient ± SD = 0.53 ± 0.12). Eyes with a visible lesion on MRI showed a significantly lower T1/T2 (mean ± SD = 0.61 ± 0.09) compared to other groups (multiple sclerosis without optic neuritis 0.74 ± 0.06, p<0.001; and healthy subjects 0.77 ± 0.09, p<0.001). Inter-eye asymmetry demonstrated a good discriminative performance, with an area under the curve of 0.83, a sensitivity of 0.74, and a specificity of 0.82. The findings indicate that conventional MRI can provide clinically meaningful information regarding optic nerve integrity and support the development of automated tools for detecting optic nerve lesions.

Interpretation of urodynamic studies (UDS) is central to diagnosing lower urinary tract dysfunction, but it is often complex, time-consuming, and subject to interobserver variability. Recent advances in artificial intelligence (AI), particularly machine learning (ML) and deep learning (DL), have enabled automated analysis of multichannel pressure-flow data and may improve objectivity and efficiency in UDS interpretation. To critically review current applications of ML and DL for interpretation of multichannel UDS data and to evaluate their performance, clinical targets, and limitations. A literature search of PubMed, Scopus, and Web of Science was performed through September 2, 2025, without language or date restrictions. A narrative review was conducted of peer-reviewed studies applying ML or DL techniques to invasive UDS signals, including vesical, abdominal, and detrusor pressures as well as urinary flow. Studies limited to uroflowmetry or non-UDS inputs were excluded. Model design, input features, validation strategies, and clinical outcomes were systematically assessed. A total of 12 studies met inclusion criteria. AI models were applied to a range of UDS tasks, including detection of detrusor overactivity, classification of bladder outlet obstruction and detrusor underactivity, multi-feature UDS pattern recognition, severity grading, and real-time event detection. Reported diagnostic performance was generally high, with many models achieving accuracies or AUCs between 80% and 95% for primary outcomes. However, study designs were heterogeneous, most datasets were retrospective and single-center, and external validation was limited. ML and DL approaches can extract clinically meaningful information from multichannel UDS recordings and demonstrate strong technical performance across several diagnostic tasks. However, variability in methodology, limited generalizability, and lack of prospective validation currently limit clinical adoption. AI-based UDS interpretation remains an emerging research area requiring standardized data, robust validation, and integration with clinical workflows.

Pathological conditions may result in clinical chemical values exceeding the linearity range of laboratory methods. In clinical practice, not all tests require exact numerical values for patient management, as useful information can still be provided when results are reported to a defined cutoff. This study describes an approach to reduce unnecessary dilutions by identifying tests that can be reported above a defined threshold. Eighty-six clinical chemistry assays on Cobas 8000 (middleware: Infinity) were reviewed at the Laboratory Medicine of the University Hospital of Padua. Fifteen analytes requiring manual dilution beyond predefined automated limits were selected. A multi-step workflow was implemented, including literature review, clinician and staff engagement, definition of upper reporting limits, middleware customisation, and retrospective and prospective performance evaluation. Key outcomes included turnaround time (TAT), auto-release rate, manual dilution frequency, estimated costs, and clinician requests for exact values. Literature review and clinician feedback supported the definition of upper reporting limits beyond which exact values were not routinely required, except for lactate dehydrogenase (LDH), where oncologic follow-up necessitated precise reporting. Implementation of the new rules eliminated most manual dilutions, substantially reducing costs and improving efficiency. P90 TAT decreased for samples above cutoff values in most analytes between 2023 and 2025. LDH was the only analyte for which manual dilution was not completely eliminated in 2025, with a residual rate of 3.38 per 1000 tests (vs. 7.32 per 1000 in 2023). Clinician feedback confirmed the appropriateness of the new reportingrules. This study shows that revising validation rules to report results above predefined thresholds provides clinically meaningful information while reducing unnecessary manual procedures. This model supports appropriate patient management and improves efficiency by enhancing process automation, consistency, and decision-making.

Abstract Gestational diabetes mellitus (GDM) is a major contributor to adverse maternal and neonatal outcomes, particularly in low-resource settings where timely diagnostic testing is not always feasible. Early identification of high-risk pregnancies using routinely collected antenatal information could support targeted screening and preventive care. This study evaluated the feasibility of early GDM risk prediction using machine learning models trained on routine antenatal care data from 3525 pregnancies in Uganda (2153 non-GDM and 1372 GDM cases). Logistic regression, Random Forest, XGBoost, and a soft-voting ensemble were developed and assessed using a held-out test set. Model performance was evaluated using receiver operating characteristic area under the curve (ROC AUC), precision–recall AUC (PR AUC), accuracy, precision, recall, and F1-score, and interpretability was examined using SHapley Additive exPlanations (SHAP). Tree-based models demonstrated strong discrimination, with Random Forest and XGBoost achieving ROC AUC values of 0.997 and PR AUC values above 0.995, while logistic regression achieved ROC AUC of 0.982. Random Forest achieved high sensitivity (recall = 0.994), missing only two GDM cases in the test set. SHAP analysis identified body mass index, high-density lipoprotein cholesterol, blood pressure, and prior metabolic history as influential predictors, with feature effects consistent with established clinical knowledge. The proposed approach is intended as an early risk-stratification and triage support tool rather than a diagnostic system. All predictors were available prior to confirmatory glucose testing, and model outputs are designed to prioritise women for further evaluation and follow-up. Although performance estimates reflect internal validation within a referral-hospital cohort, the findings demonstrate that routinely collected antenatal variables contain clinically meaningful predictive information. With external and prospective validation, interpretable machine learning models may support more efficient screening and monitoring of gestational diabetes in resource-constrained antenatal care settings.

Optimising clinical information extraction: a comparative study of retrieval-augmented generation techniques in clinical notes.

Tau protein aggregation in the brain is a hallmark of Alzheimer's disease (AD). Positron emission tomography (PET) is the only in vivo method to visualize tau pathology and estimate both its burden and regional distribution, but the use of tau-PET is constrained by high cost and limited accessibility. Here, we develop a deep learning model to synthesize tau-PET scans from more accessible data: structural magnetic resonance imaging (MRI), demographics, and when available, blood biomarkers. We included 5,191 participants across the AD continuum or with another neurological disorder from 13 cohorts (mean age 70 years, 51% female) and optimized a 3D U-Net neural network with residual and attention units for this task. In held-out test data, synthetic tau-PET reliably modeled tau burden, with correlations of R=0.77-0.86 with true tau-PET across individuals in common AD regions of interest. Spatial similarity between synthetic and true tau-PET was likewise high, with mean regional correlation of R=0.75. Synthetic scans also captured clinically meaningful prognostic information comparable to true tau-PET, including distinction between early (HR=12, p<0.001) and late (HR=45, p<0.001) stages of tau accumulation. These findings demonstrate that clinically informative synthetic tau-PET scans can be generated from widely available modalities using deep learning, potentially offering a scalable and cost-effective approach for estimating tau AD pathology in the brain.

Early identification of gram-negative bacteremia in intensive care units (ICUs) remains challenging at the time of blood culture sampling, when clinical signs are often nonspecific and existing diagnostic approaches typically rely on single-timepoint measurements. We conducted a retrospective cohort study of adult ICU patients admitted between July 2022 and January 2024 to investigate whether short-term longitudinal patterns in routinely collected clinical and laboratory data contain diagnostically relevant information for gram-negative bacteremia. Clinical and laboratory variables were extracted at three consecutive timepoints (Day -2, -1, and 0 relative to blood culture collection), and diagnostic models incorporating this temporal information were developed using complementary statistical and machine-learning approaches. Model performance was evaluated on a held-out test set using discrimination, calibration, and decision curve analysis. Among 568 patients, models incorporating short-term longitudinal data demonstrated good and consistent discrimination for gram-negative bacteremia (AUC range 0.81-0.83) with good calibration after recalibration. Diagnostic performance was stable across modeling approaches, indicating robustness of the underlying signal rather than dependence on a specific algorithm. Decision curve analysis suggested higher net benefit for model-based risk stratification compared with treat-all or treat-none strategies across clinically relevant threshold probabilities. Hemoglobin, creatinine, and albumin consistently emerged as influential contributors. These findings indicate that short-term longitudinal clinical trajectories contain diagnostically meaningful information for gram-negative bacteremia at the time of blood culture sampling and support further external validation and prospective evaluation prior to clinical implementation.

To assess the concordance between blood culture isolates and intraoperative deep tissue cultures in patients with confirmed fracture-related infection (FRI). Retrospective Cohort Study. Academic Medical Center. This Institutional Review Board-approved study included patients 18 years and older diagnosed with a confirmed FRI according to the FRI Consensus Group criteria who, at time of irrigation and debridement (I&D), underwent deep tissue culture (TC) as well as concurrent blood culture (BC) testing (in the Emergency Department or inpatient setting). The decision to perform BC testing was left to the discretion of the initial treating providers at the time of this presentation. Microbiological data were reviewed from the electronic medical record. Infections were classified as monomicrobial (either gram-positive or gram-negative), polymicrobial, or culture negative. Pathogen concordance between blood and intraoperative tissue cultures was analyzed. 84 patients were included with both intraoperative deep TC and concurrent BC. This cohort had a mean age of 56.2 ± 20.3 years and consisted of 33 females (39.3%). BC were never ordered by the orthopedic surgeon. Microbial analysis of deep tissue specimens identified 29 gram-positive infections, 18 gram-negative infections, 33 polymicrobial infections, and 4 culture-negative cases. Of the 84 BC analyzed, 69 (82.1%) were culture-negative and 15 (17.9%) were culture-positive. BC results were discordant with their respective TC isolates in 76 of 84 (90.4%) cases. This discordance in 76 cases was driven by negative BC in the setting of positive TC (69/76, 90.8%), while a smaller proportion reflected growth of different organisms in BC compared to TC (7/76, 9.2%). Concordance was observed in only 8 of 84 (9.6%) cases, in which BC identified at least one pathogen sampled from TC. BC yielded negative culture results 17 times as often as TC. McNemar's test revealed a highly significant difference in culture-positivity rates (χ2=65, p<0.0001), while Cohen's Kappa for agreement was 0.022, indicating minimal agreement between BC and TC results. These results suggest that blood cultures were part of some workflows for patients presenting with infections, but they did not reflect the true bony pathogens nor contribute meaningful diagnostic information in most cases of confirmed fracture-related infection (FRI) according to the FRI Consensus Group criteria. While blood culture testing is important in the evaluation of systemic infection from, it does not provide orthopedic surgeons with information that informs the management or treatment of the FRI itself. III.

Exposure and response prevention (EX/RP) is a first-line psychological treatment for obsessive-compulsive disorder (OCD), yet treatment response varies across individuals. Given the substantial heterogeneity in OCD symptom presentations, examining changes in obsessive-compulsive (O-C) symptom profiles, rather than overall symptom severity, may clarify why patients respond differently to treatment. Accordingly, this study aimed to identify O-C symptom profile classes across treatment and to characterize transitions between classes during EX/RP. The sample included 152 patients with OCD from two clinical trials that delivered 17 twice-weekly EX/RP sessions over 8 weeks, with assessments conducted at baseline, midpoint and post-treatment. Latent profile analysis was used to identify O-C symptom profiles at each time point, followed by latent transition analysis to examine how individuals transitioned between profile classes over time. Distinct O-C profiles were identified at each time point, reflecting varying severity across Obsessive-Compulsive Inventory-Reviseddimensions. Most patients maintained stable profiles with reduced overall symptom severity across treatment, supporting EX/RP's effectiveness. However, some profiles-particularly those characterized by elevated neutralizing symptoms-showed more variable transitions. A novel profile emerged at midpoint, with lower baseline avoidance predicting the membership. These findings suggest that meaningful clinical information is gained by monitoring changes in O-C symptom profiles alongside overall severity during EX/RP. In particular, elevated neutralizing symptoms may signal a more variable response, whereas hoarding elevations may warrant early identification and proactive planning for targeted strategies beyond standard EX/RP. Incorporating profile-based monitoring into clinical practice may help clinicians more precisely tailor EX/RP to maximize therapeutic benefit.

Statistical analysis of in vitro assay data is a critical step towards a good interpretation of biological responses. However, it is still frequently undermined due to inappropriate statistical procedures, misinterpretation, insufficient statistical power-often resulting from small sample sizes-or poorly defined methodologies, in case of standardized tests. In line with practices commonly adopted in clinical studies and more recently in biomonitoring research, the use of thresholds for interpreting results may improve the robustness of conclusions. This work presents the application of a methodology for defining thresholds using a normal distribution-based approach. As a case study, these thresholds were applied to analyze data obtained from the DLES test (Dicentrarchus labrax estrogen screen test), an in vitro screening tool designed to detect interactions between chemicals and nuclear estrogen receptors in D. labrax. The results were subsequently compared with methods derived from the OECD TG 455, as well as with non-parametric statistical analyses. By applying normal distribution-based thresholds, data analysis was simplified and the reliability of the DLES test results was increased, especially when compared with hypothesis tests. Also, this was especially true when studying non-model species, for which standard reference substances are rarely available. However, special attention should be paid to the size of the initial dataset used to define the thresholds. The methodology implemented here could provide insight for other in vitro assays. Overall, this article encourages the reflection on approaches to in vitro data analysis.

The traditional immunogenicity paradigm resulting in anti-drug antibody (ADA) and neutralizing antibody (NAb) positive/negative status and ambiguous or imprecise titers may overlook clinically relevant effects of ADA on pharmacokinetics (PK) and pharmacodynamics (PD). Employing risk-based strategies that integrate PK and PD analyses with ADA magnitude using signal-to-noise (S/N) can provide early insight into potential clinical impact of immunogenicity. In a Phase 1 evaluation of DLX-2323, a humanized single-chain variable fragment (scFv) antibody that binds human IL-1β, ADA-sensitive PK and PD assays were employed to measure DLX-2323 concentration, PD activity, and assess the impact of ADA on PK and PD in healthy participants. The presence of pre-existing ADA was observed in one-third of participants and resulted in high variability of DLX-2323 exposure and PD activity. Pre-existing ADA magnitude correlated with a drug-sustaining impact on PK and PD, with lower clearance (CL/F) and volume of distribution (Vd/F) of DLX-2323 relative to increasing ADA signal. Based on the immunogenicity risk assessment, the use of ADA-sensitive free PK and PD assays to measure in vivo ADA effects on drug exposure and PD activity provided clinically meaningful information that standalone neutralizing antibody assays could not.