Development of a smartphone-based bone maturity classification algorithm with XAI for beef carcass grading

Youth hospitalized for a mental health (MH) condition frequently experience MH-related avoidable hospital days (MH-ADs), or days in which they remain hospitalized for a MH reason despite not requiring services unique to a medical hospital. However, there is currently no reliable method for identifying MH-ADs across healthcare systems, preventing investigation of this systemic problem. A universal and efficient method of determining MH-ADs is needed to guide improvements in access to care for youth with MH needs while reducing unnecessary medical hospital days. Our objectives were to create an administrative algorithm for identifying MH-ADs and to evaluate the algorithm's validity using clinical data from a single hospital site. The resulting algorithm, drawing on coding and billing data from the Pediatric Health Information System Database, identified MH-ADs with good sensitivity (79.9%), specificity (79.2%), and positive predictive value (95.1%), but low negative predictive value (43.5%) when compared to clinically determined MH-ADs.

Development of a volumetric PTV algorithm incorporating stereoscopic shadowgraphy for measuring turbulent and complex vortical flow patterns

Introduction & Purpose: Ultrasonography is widely used to assess skeletal muscle and tendon properties, such as architecture, cross-sectional area, and tissue stiffness (Sarto et al., 2021). Despite its growing application in different scenarios, and the increasing call for open data access and sharing in clinical research, there remains a significant scarcity of public datasets in this field. This lack of accessible and standardized public datasets limits large-scale analysis algorithm studies, trainee training and the development of image analysis algorithms. To address this, we developed the Universal Musculoskeletal Ultrasonography Database (UMUD), a web application designed to facilitate access to these datasets and foster standardization and innovation in musculoskeletal ultrasonography imaging research. Methods: UMUD is an online repository that aggregates and indexes metadata from publicly available musculoskeletal ultrasonography datasets hosted on platforms like the Open Science Framework and Zenodo. The web application (https://universalmuscledatabase.streamlit.app/) is built using a streamlit (v1.35.0) frontend and mongoDB for its database infrastructure. Standardized metadata descriptors, i.e., muscle group, ultrasound device, participant demographics, are implemented using a combination of pydantic models (v1.10.0) and json schemata to ensure reproducibility and ease of usage for contributing data. UMUD provides detailed instructions for community contributions, including tools for data anonymization. Results: Currently, UMUD hosts 11 datasets from 10 studies, comprising 75,569 images and 2,573 videos from 1,769 participants. The database covers nine lower-limb muscles and one muscle–tendon junction (distal triceps surae and Achilles tendon), captured using various modalities including static imaging, dynamic video, and 3D reconstructions. Datasets include measurements across proximal, middle, and distal regions of each muscle. Benchmark datasets are provided for trainee training and algorithm evaluation, including multi-expert annotated images, fascicle and cross-sectional area overlays, and fully labeled datasets for deep-learning model training. Additionally, UMUD lists available automated image analysis algorithms as a reference for community use. Discussion: UMUD provides an initial foundation for open, standardized, and community-driven musculoskeletal ultrasound research. By centralizing datasets and metadata, it facilitates reproducible research, algorithm benchmarking, and operator education. The inclusion of multi-expert and labeled benchmark datasets supports both training and the development of automated analysis methods. Future directions include expanding dataset coverage, enhancing interactive visualization tools, and launching community challenges for algorithm benchmarking to accelerate innovation in the field. Conclusions: In conclusion, UMUD addresses relevant challenges in musculoskeletal ultrasonography by providing a centralized, standardized repository of datasets and tools. It promotes transparency and innovation in the field, supporting reproducible research and advancements in automated image analysis. Future developments include adding datasets, expanding functionalities and introducing community-driven algorithm development challenges.

The appearance of whole slide biopsy images is greatly affected by various factors such as laboratory procedures or the choice of digital slide scanners. The resulting variations in image styles within and across batches of histological images represent one of the major obstacles to the development of generalizable machine learning algorithms. To overcome this challenge, a lot of research has focused on stain normalization and stain augmentation techniques. While such approaches provide effective strategies to reduce stain variation or increase stain invariance, respectively, they typically involve only limited modelling or sampling of the underlying stain style distribution. Tools for a streamlined sampling of different aspects of such a distribution, which would be crucial e.g. for explicitly evaluating machine learning robustness across or with respect to major stain styles, remain largely missing. Here, we present the StainStyleSampler, a toolkit for (i) the exploration and modelling of stain style variations, and (ii) the automated sampling of images or styles capturing the core components of this variation. The tool enables the extraction of various colour features and deconvolved stain components, visualization of such features directly or after dimensionality reduction, modelling of style distributions using binning, clustering, and density mapping, and automated sampling of the most representative reference images. We believe that this software will equip pathologists and computer-scientists with a more versatile set of tools that can substantially aid in both the exploration and sampling of stain variation across whole slide images.

Significant developments made in quantum hardware and error correction recently have been driving quantum computing toward practical utility. However, gaps remain between abstract quantum algorithmic development and practical applications in computational sciences. In this perspective article, we propose several properties that scalable quantum computational science methods should possess. We further discuss how block-encodings and polynomial transformations can potentially serve as a unified framework with the desired properties. Recent advancements on these topics are presented, including the construction and assembly of block-encodings, and various generalizations of quantum signal processing (QSP) algorithms to perform polynomial transformations. The scalability of QSP methods on parallel and distributed quantum architectures is also highlighted. Promising applications in simulation and observable estimation in chemistry, physics, and optimization problems are presented. We hope this perspective serves as a gentle introduction to state-of-the-art quantum algorithms for the computational science community and inspires future development of scalable quantum computational science methodologies that bridge theory and practice.

Accurately predicting ovarian response and determining the optimal starting dose of follicle-stimulating hormone (FSH) remain critical yet challenging for effective ovarian stimulation. Currently, there is a lack of a comprehensive model capable of simultaneously forecasting the number of oocytes retrieved (NOR) and assessing the risk of early-onset moderate-to-severe ovarian hyperstimulation syndrome (OHSS). This study aimed to establish an integrated mode capable of forecasting the NOR and assessing the risk of early-onset moderate-to-severe OHSS across varying starting doses of FSH. This prognostic study included patients undergoing their first ovarian stimulation cycles at 2 independent in vitro fertilization clinics. Automated classifiers were used for variable selection. Machine learning models (11 for NOR and 11 for OHSS) were developed and validated using internal (n=6401) and external (n=3805) datasets. Shapley additive explanation was applied for variable interpretation. The best-performing models were incorporated into a web-based prediction tool. For NOR prediction, 17 variables were selected, with the gradient boosting regressor achieving the highest performance (internal dataset: R2=0.7978; external dataset: R2=0.7924). For OHSS prediction, 19 variables were identified, and the LightGBM model demonstrated superior performance (internal dataset: area under the receiver operating characteristic curve=0.7588; external dataset: area under the receiver operating characteristic curve=0.7287). Shapley additive explanation analysis highlighted the FSH starting dose to BMI ratio and baseline antral follicle count as key predictors for NOR and OHSS, respectively. Dose-response curves were generated to visualize predicted outcomes with varying FSH starting doses. The models were implemented in a user-friendly, research-oriented online prototype, individualized ovarian stimulation guide (InOvaSGuide). This study introduces an integrated framework for predicting NOR and early-onset moderate-to-severe OHSS risk across different FSH doses. Future prospective evaluation is needed before clinical implementation.

BackgroundCOVID-19 forecasting models have been used to inform decision-making around resource allocation and intervention decisions, such as hospital beds or stay-at-home orders. State-of-the-art forecasting models often use multimodal data, including mobility or sociodemographic data, to enhance COVID-19 case prediction models. Nevertheless, related work has revealed under-reporting bias in COVID-19 cases as well as sampling bias in mobility data for certain minority racial and ethnic groups, which affects the fairness of COVID-19 predictions across racial and ethnic groups.ObjectiveThis study aims to introduce a fairness correction method that works for forecasting COVID-19 cases at an aggregate geographic level.MethodsWe use hard and soft error parity analyses on existing fairness frameworks and demonstrate that our proposed method, Demographic Optimization (DemOpts), performs better in both scenarios.ResultsWe first demonstrate that state-of-the-art COVID-19 deep learning models produce mean prediction errors that are significantly different across racial and ethnic groups at larger geographic scales. We then propose a novel debiasing method, DemOpts, to increase the fairness of deep learning–based forecasting models trained on potentially biased datasets. Our results show that DemOpts can achieve better error parity than other state-of-the-art debiasing approaches, thus effectively reducing the differences in the mean error distributions across racial and ethnic groups.ConclusionsWe introduce DemOpts, which reduces error parity differences compared with other approaches and generates fairer forecasting models compared with other approaches in the literature.

Peripheral intravenous infiltration and extravasation injuries (PIVIEs) in pediatric patients are a significant yet underrecognized source of preventable harm, often due to inconsistent detection and limited standardization. This study implemented a structured quality improvement initiative to enhance early detection, classification, and management of PIVIEs at C.S. Mott Children's Hospital, supporting the institution's goal of zero preventable harm. This single-center, prospective quality improvement study was conducted at a 347-bed academic pediatric hospital from January 2023 to December 2024. All patients with a documented PIVIE were included. Key interventions included creation of a multidisciplinary PIVIE Prevention Task Force, standardized event review, development of a pediatric infusion agent risk classification algorithm, and updated antidote protocols. Outcomes measured included PIVIE reporting volume, antidote timeliness, and classification documentation. Descriptive statistics were used to assess trends. Following intervention, 628 PIVIEs were reported-a 465.8% increase from 111 events reported during the 2021-2022 period. Implementation of an institutional standard of care form and a centralized database improved documentation and expedited antidote administration. A pediatric-specific classification system was developed and applied to over 90 high-use agents, enabling risk-based clinical decision-making. Interdisciplinary, structured interventions markedly improved reporting, documentation, and treatment of pediatric PIVIEs. Adoption of standardized classification tools and review processes may help hospitals reduce IV-related complications. Policymakers and accrediting bodies should consider incentivizing pediatric vascular safety metrics, supporting integration of pediatric-specific risk frameworks, and encouraging cross-institutional data sharing to drive broader improvements in care.

This paper presents a simplified hybrid modulation method for operating dual-active-bridge (DAB) converters that power inverters by integrating single-phase shift (SPS) and triple-phase shift (TPS) modulation schemes. It covers the design and control algorithm development, performance analysis, as well as highlights its benefits and limitations. While full TPS implementation is highly complex, this work selects a specific TPS operating mode to enhance DAB converter efficiency in low-power conditions with minimal control effort. In the proposed method, a hysteresis controller is employed to regulate the DAB modulation at a defined power threshold. This poses a significant challenge, especially when a single proportional-integral (PI) controller is employed to regulate output power with a minimal set of control parameters applicable to both modulation schemes. Moreover, in addressing these challenges, a trade-off between high efficiency and fast dynamic response is also considered, with greater emphasis placed on efficiency when developing the controllers. Meanwhile, the inverter output voltage is independently controlled regardless of the DAB operation to further simplify the overall process. Small signal modeling and closed-loop control of DAB-based inverter with the proposed hybrid modulation are also presented. Its functionality and performance have been verified through simulation and a developed small-scale DAB-based inverter prototype.

Design and Development of Early Prediction Algorithm for Ventricular Fibrillation Using Novel Hybrid Deep Learning Model

To determine how certolizumab pegol (CZP) dose and dose adjustments influence CZP plasma trough levels to facilitate therapeutic drug monitoring of CZP. The effect of CZP dose and dose adjustments on CZP plasma trough levels was evaluated post hoc using longitudinal data from a 52-week randomized phase III trial (RAPID 1) and its open-label extension trial. Patients with active rheumatoid arthritis treated with methotrexate for ≥6 months were randomized to CZP 200 mg, 400 mg, or placebo every other week (EOW). Patients in the extension trial were initially treated with CZP 400 mg EOW, then reduced to 200 mg EOW after ≥6 months. Of 982 randomized patients, 846 patients entered the open-label extension trial. Median (interquartile range) plasma CZP concentrations after 12 weeks of treatment were 21.3 mg/L (14.7, 27.7) in the 200-mg group and 38.3 mg/L (29.2, 63.8) in the 400-mg group and increased from 18.3 (12.4, 26.5) to 43.4 (26.8, 63.3) mg/L after dose escalation from 200 to 400 mg EOW. Following CZP dose reduction from 400 mg to 200 mg, median CZP levels decreased from 36.1 (24.9, 49.0) to 17.2 (11.5, 23.1) mg/L. CZP plasma concentrations were influenced by both dose and dose adjustment in a predictable manner, with median plasma levels twice as high in the 400-mg group than in the 200-mg group, with a 2-fold increase after the dose increase from 200 to 400 mg. This facilitates the development of algorithms for therapeutic drug monitoring of CZP.

Integration of quantum artificial intelligence in disease diagnosis: A review of methods and applications.

Energy-based devices (EBDs) are increasingly used to manage acne and its sequelae. While literature supports the use of appropriate skin care for acne, few studies address how to effectively integrate skincare with EBDs. Six dermatologists from North America, participated in a live meeting to develop an acne treatment algorithm integrating skincare and EBDs. Six additional advisors contributed through a pre-meeting survey (along with 94 other physicians). The eleven dermatologists (authors) from Asia, Europe, Australia, and North America participated in algorithm development and manuscript review. The proposed algorithm describes how to integrate skin care with the use of EBDs in clinical practice. This algorithm provides an approach for managing acne and best practices for integrating skin care with EBDs when treating acne and acne sequelae. &nbsp.

High spatial resolution GLASS FAPAR (version 2) product from Landsat imagery: Algorithm development using a knowledge transfer strategy

Motion-induced electromagnetic interference remains a major obstacle to the accurate interpretation of surface-recorded biosignals collected during movement. This study presented a physics-based rigid-body model that integrated electromagnetic theory with a kinematic framework to describe the generation of motion-induced artifacts in surface biosignals through electromagnetic induction. The model was derived from Faraday's law and a 6D rigid-body kinematic formulation, which coupled rotational and translational motion to spatial magnetic-field gradients and curvature. This formulation predicted that any conductive loop moving within a nonuniform magnetic field produced a time-varying electromotive force (EMF) determined by the interaction between motion, field geometry, and sensor orientation. To illustrate and validate the theoretical model, computational simulations reproduced treadmill locomotion under two conditions: (1) an idealized fixed-cadence case with time-invariant field gradients, and (2) a realistic varying-cadence case incorporating stride-to-stride jitter and event-related spectral perturbation baseline correction. The simulated EMF spectra exhibited motion-locked harmonic patterns extending up to 15 Hz with electrode-dependent variations in magnitude and broadened harmonic envelopes, closely matching empirical treadmill electroencephalography spectra. Accelerometer spectra displayed broader harmonic content up to 50 Hz, consistent with their direct measurement of kinematic oscillations. Quantitative decomposition further revealed that rotational motion dominated the induced EMF, with smaller, electrode-dependent contributions from translation. Robustness analyses indicated that dominant harmonic structure is preserved under multi-axis kinematics and increased magnetic-field complexity, with greater sensitivity confined to weaker higher-order components. These results demonstrated that harmonic contamination could emerge naturally from rigid-body motion in a spatially varying magnetic field, providing a physics-based foundation for interpreting motion artifacts in surface electrical potentials and motivating practical mitigation strategies that incorporate motion and magnetic-field measurements. Through principled understanding and physics-based modeling of motion-induced electromagnetic artifacts, this framework supports interpretation of surface biosignals during movement and motivates the development of mitigation algorithms.

Pressure injury surveillance in the intensive care unit: Development, validation, and clinical application of a natural language processing algorithm.

Development and testing of a new pore network algorithm for modeling flows of power law fluids in porous media

New algorithm development for real-time dynamic simulation of thermal–hydraulic grids and artificial intelligence-assisted hierarchical control of integrated concentrated solar power, steam Rankine cycle, and high-temperature steam electrolysis systems

Development of nanoscale magnetic and plasmonic materials for applications such as Heat Assisted Magnetic Recording requires precise control and understanding of the materials’ microstructure. Scanning Electron Microscopy (SEM) has the speed and resolution to characterize grain structure with high throughput, but there is little precedence for imaging sub-50 nm, crystallographically-aligned metal grains in a traditional SEM without specialized detectors or optics. Imaging ungrounded micron-scale metal structures on wafers presents further challenges due to charging. By optimizing imaging parameters for each sample, sub-50 nm and sub-25 nm metal grains were captured. Monte Carlo simulations were used to understand the depth of backscattered electron signal for the stacks of materials and the effect of grain boundary tilt on grain boundary contrast. High-throughput SEM grain imaging as demonstrated in this work yields large materials characterization datasets without expensive detectors or specialized hardware. Translating the qualitative SEM grain images to quantitative characterization requires continued algorithm development, yet there are significant opportunities for automated materials development and structure–property elucidation for SEM grain imaging combined with computer vision. The present and future of magnetic devices requires nanoscale materials control, and high-throughput SEM grain imaging is a promising metrology route for understanding and producing those structures.