Mind the Gap: Addressing Missing Person Time When Estimating Outcome Incidence in Longitudinal Data.

Cross-sensor spectral fusion for coastal water physicochemical assessment using optimized ensemble machine learning approaches.

Abstract Accurate assessment of solar resources is critical for photovoltaic (PV) project feasibility and energy auctions, yet satellite-based estimates can differ significantly from ground measurements, creating uncertainty and motivating the need for reliable site adaptation methods. Within this context, the present article proposes a method for enhancing the linear regression site adaptation method by incorporating solar irradiance band separation, clear-sky classification index (Kc) analysis, and machine learning algorithms (artificial neural network (ANN)) while also evaluating the influence of irradiance data collection period and diverse climatic conditions validation. Results show that relative root mean square error (rRMSE) improvements were site dependent: all methods improved rRMSE at sites with over 75% clear-sky days, while only ANN was effective at 50% clear-sky sites. For Florianópolis (Brazil), single-year analyses showed that the irradiance band method achieved the lowest relative mean bias error (rMBE) in 50% of cases, compared to 33% for linear regression and 17% for ANN. The findings suggest that the proposed solar irradiance band classification is particularly well-suited for regions with stable and abundant solar resources. Additionally, results indicate that using two years of measured data would offer significant improvements over a one-year period and could reduce uncertainties for long-term PV plant performance assessment and aid implementation by industry planners and researchers in both government and nongovernment organizations.

Abstract Understanding interactions between incoming shortwave (SW) solar radiation and clouds is essential for quantifying and modeling Earth's Radiation Budget (ERB). Ice clouds are particularly problematic due to their wide variability in crystal habits, sizes, and shapes. In this study, data from NASA's Cloud and Earth Radiative Energy System (CERES) are used to identify single‐layer overcast ice clouds and calculate surface and top‐of‐atmosphere (TOA) SW fluxes using the Canadian Centre for Climate Modeling and Analysis (CCCma) Radiative Transfer Model (RTM). A total of 361 SW flux observations from 11 surface sites spanning different climatic regions, together with CERES SYN1deg satellite observations at the TOA, are used to evaluate the CCCma RTM's performance. The CCCma RTM exhibits mean bias errors (MBEs) of +3.7 W m −2 at the surface and +4.1 W m −2 at the TOA, with root mean square errors (RMSEs) of 72.7 and 33.2 W m −2 , respectively. Correspondingly, the CERES SYN1deg Fu‐Liou RTM shows MBEs of −12.1 and +18.5 W m −2 and RMSEs of 75.0 and 34.5 W m −2 for surface and TOA, respectively. MBE differences between the two RTMs are due to differing treatments of model physics, while their larger RMSEs at the surface result from both imprecise inputs and spatial variabilities of both inputs and surface observed flux.

Abstract The spatio-temporal distribution of winter precipitation over the Arabian Peninsula (AP) is crucial for managing various socioeconomic sectors. Due to limited observational data, high-resolution atmospheric models are often used to investigate rainfall distribution across the region. However, uncertainties in current models are strongly influenced by the representation of clouds, moist convection, and complex topography. Reducing the grid spacing to a few kilometers using a cloud-resolving model (CRM) allows for improved treatment of clouds and associated hydro-climatological processes, which can help reduce uncertainties in model precipitation physics. To address this, we conducted multi-year CRM simulations using the WRF model at 2 km horizontal resolution, specifically targeting the winter season, to simulate seasonal precipitation patterns over the AP from 2006 to 2016. These CRM outputs are validated against available in-situ , gridded, remotely sensed observations and reanalyses fields. Furthermore, we used ERA5 reanalysis fields to evaluate circulation, thermodynamic and microphysical processes in the CRM. Our results demonstrate that CRM simulated rainfall agrees well with the observed rainfall patterns, albeit for some wet bias (ranging between 0.83–1.88) and root mean square error (0.81–1.40) over mountainous regions. Precipitation statistics confirm that the CRM adequately captures the spatial extent and variability of winter rainfall over the AP. The model also reveals that moisture transport from the Red Sea, Mediterranean Sea, and Arabian Sea substantially influences the regional precipitation patterns. Notably, the CRM enhances the simulation of extreme precipitation events, accurately capturing their spatial distribution, intensity, and frequency. We also examined the underlying physical mechanisms driving these precipitation dynamics using the CRM simulations. Overall, the findings demonstrate that the CRM provides a more realistic representation of fine-scale precipitation features and the associated physical processes across the AP. This study offers valuable insights into the application of high-resolution CRM frameworks for improving the prediction of precipitation extremes in this arid region.

In the Arctic region, a decline in ice and snow cover has been observed in recent years, resulting in adverse effects on the climate, hydrological events, biological processes, and human populations. Monitoring changes in ice and snow cover using satellite imagery or models is common, while novel research is beginning to use ground-based camera systems for in situ monitoring of cryosphere elements. This study focused on maximizing the usage of ground-based time-lapse imagery from trail cameras for ice and snow studies from 2016-2022 within the context of the changing Arctic climate by monitoring lake ice and snow at five lakes (Resolute Lake, Small Lake, North Lake, Plateau Lake, and Hunting Camp Lake) near Resolute and Nanuit Itillinga, Nunavut, in the Central Canadian High Arctic. A semi-automated technique using image classification tools was developed to quantify the progression of ice and snow extent in the camera view. The image classification yielded an overall classification accuracy of 86%, and a Kappa coefficient of 0.79 from nearly 13 000 images, indicating a strong and viable monitoring system despite some variance in performance from viewing conditions. Lake ice and snow phenology dates determined from classified imagery had averages generally within a few days of observations (mean bias error of 2-9 days). Average ice duration was 308 days (September 20 to July 25), and average snow duration was 298 days (September 14 to July 8). The camera-based data extraction technique is a viable tool, not only for tracking long-term changes in snow/ice conditions, but also for validating satellite or modelling work in other logistically challenging environments. Thus, this methodology to monitor Arctic ice and snow phenology can support better projections for future responses to climate change.

Abstract Behavioural finance has traditionally conceptualised investor behaviour through the lens of cognitive bias and decision error. While valuable, this perspective underexamines the structured and relational role of emotion within professional financial advice. This study moves beyond behavioural bias by developing a process-based taxonomy of client emotional expression grounded in longitudinal field evidence. Drawing on 1,236 recorded client interactions collected over a four-year period within a defined Australian financial planning practice, the research systematically classifies emotional expressions across four stages of the advice process: Discovery, Strategy Development, Implementation, and Review. Using a structured qualitative coding framework with inter-rater reliability safeguards and blinded assessment procedures, the study identifies recurring primary, secondary, and interactive emotional categories and maps their distribution across the advice lifecycle. Findings demonstrate that client emotions are not random or episodic; rather, they cluster predictably according to process stage and often follow identifiable transition pathways (e.g., anxiety to trust, confusion to relief). The analysis further reveals the presence of relational emotions - such as conditional trust, reassurance, and identity affirmation - that are co-constructed within adviser-client interaction and are not adequately captured in traditional behavioural finance models. By developing a structured taxonomy of emotional expression specific to financial planning, this research advances the vocabulary and classification of emotion in applied financial contexts. The resulting framework provides a replicable foundation for future empirical testing, informs emotional intelligence training for advisers, and offers a basis for process-sensitive engagement strategies that strengthen trust formation, decision quality, and long-term adherence to financial plans. In doing so, the study positions financial advice not merely as a technical decision-making exercise, but as a staged emotional regulation process integral to durable client outcomes.

Soil total nitrogen (STN) is a crucial indicator of crop productivity and soil health. Accurate monitoring of STN is essential for optimizing nitrogen management and achieving sustainable agricultural development. An adequate and timely STN supply serves as a key physiological basis for promoting effective tillering, flower stalk development, and continuous multibatch bud formation in Hemerocallis citrina Baroni. To address the challenges posed by the high-dimensionality of hyperspectral data and the dynamic spectral response of STN across different growth stages, this study employed spectral resampling to select feature bands based on Sentinel-2 sensor data(Simulation of Sentinel-2 Bands, SSB method). Specifically, hyperspectral data were collected under laboratory controlled conditions (constant temperature darkroom, standard light source, air-dried ground soil), simulated Sentinel-2 sensor bands through spectral resampling (SSB method), and constructed an STN prediction framework based on 8 machine learning algorithms(random forest, extreme gradient boosting, back propagation neural network (BPNN), genetic algorithm-optimized BPNN (GA-BPNN), convolutional neural networks (CNN), and a hybrid CNN-bidirectional long short-term memory-attention model). The model performance was comprehensively evaluated using the coefficient of determination (R 2 ), root mean square error (RMSE), mean absolute error (MAE), and mean bias error (MBE). This study aims to establish laboratory-scale soil-spectral chemical relationship baselines, providing band selection and algorithm validation references for subsequent field remote sensing applications, rather than directly developing field operational systems. The results showed that: (1) the three-band spectral index TBI3 exhibited the highest correlation with STN across the full growth period (R=0.7354). The optimal indices for specific growth stages were TBI4, TBI3, and TBI5 for the spring seedling/leaf expansion, bolting/flowering, and bud emergence stages, respectively, with TBI-series indices exhibiting significantly superior performance compared to two-dimensional indices; (2) the GA-BPNN model achieved the highest accuracy for the full growth period, with a test R 2 of 0.6284, along with the lowest MAE (0.0693 g·kg -1 ) and RMSE (0.0879 g·kg -1 ), demonstrating outstanding generalization capability; and (3) the GA-BPNN model outperformed the other models in comparative analyses across different growth stages, and the growth stage-specific integrated modeling method showed higher prediction accuracy and enhanced resistance to overfitting (both training and test R 2 exceeded 0.6, with the gap reduced to 0.0064). Based on these findings, we propose a technical framework termed "SSB-SPXY-GA-BPNN-growth stage adaptation", which provides theoretical and methodological support for precise STN monitoring and variable-rate fertilization.

This study investigates a range of parameter estimation methods for the Half-Logistic Inverse Rayleigh Distribution (HLIRD) under two distinct sampling frameworks: ranked set sampling (RSS) and simple random sampling (SRS). The estimation techniques considered include maximum likelihood estimation, ordinary and weighted least squares, and the maximum and minimum product of spacings methods. Model adequacy is evaluated using five goodness-of-fit criteria: the Anderson–Darling (AD) statistic, its right- and left-tail variants, the second-order left-tail AD statistic, and the Cramér–von Mises statistic. An extensive simulation study is conducted to thoroughly evaluate and compare the performance of the proposed estimators while maintaining a fixed total number of observations across both sampling schemes. The practical relevance of the proposed methods is further illustrated through an application to a real dataset consisting of 69 carbon fiber specimens, with tensile strength measurements (in GPa) recorded at a gauge length of 20 mm. The numerical results demonstrate that estimators based on RSS consistently outperform their SRS counterparts across all considered performance measures, including mean squared error, bias, and mean absolute relative error. Overall, the findings highlight the advantages of employing RSS for parameter estimation of the HLIRD, particularly due to its superior efficiency in small-sample scenarios.

Systematic bias between therapeutic drug monitoring assays may lead to inappropriate treatment decisions in clinical practice. While such bias is well recognized, its impact on model-informed precision dosing remains unexplored. In this study, we evaluate how assay bias affects the predictive performance of population pharmacokinetic models, using ustekinumab in patients with Crohn's disease as an example. We repurposed data from 83 patients with Crohn's disease. Ustekinumab concentrations were measured using both an homogeneous mobility shift assay and enzyme-linked immunosorbent assay. Two corresponding population pharmacokinetic models were developed. Bayesian forecasting was performed under matched and mismatched combinations of assay data and population pharmacokinetic models. Predictive accuracy and precision were assessed using relative bias and relative root mean square error, with predefined thresholds for clinical acceptability. Agreement between assays and clearance estimates was evaluated using Bland-Altman plots, Deming regression, and concordance correlation coefficients. Model prior flattening strategies were explored to mitigate mismatches between model priors and therapeutic drug monitoring data. Ustekinumab concentrations measured by the homogenous mobility shift assay were overall 8.1mg/L higher than those measured by an enzyme-linked immunosorbent assay (95%confidence interval -23.6, 39.7). Clearance estimates from the homogenous mobility shift assay-based population pharmacokinetic model were systematically lower (0.107L/day; relative standard error, 7.6%) compared with those from the enzyme-linked immunosorbent assay-based population pharmacokinetic model (0.235L/day; relative standard error, 5.4%). When assay data and population pharmacokinetic models were matched, Bayesian forecasting yielded clinically acceptable predictions across all scenarios (relative bias <20%, 95% confidence interval including zero). Mismatched combinations led to reduced accuracy. Precision was highest using the homogenous mobility shift assay data, irrespective of the population pharmacokinetic model. Flattening strategies improved predictive performances in some mismatched scenarios but did not fully recover bias. Assay bias has a clinically relevant impact on the predictive performance of model-informed precision dosing. Our findings underscore the importance of aligning the therapeutic drug monitoring assay format with the assay format used to build the population pharmacokinetic model to ensure accurate and clinically acceptable dosing predictions.

Assessing position sense in motion: Reliability of a dynamic joint position reproduction test.

Estimators are essential tools in survey sampling, enabling the approximation of unknown population parameters from limited sample data. Full population surveys are often impractical due to cost, time, and logistical constraints, making estimators such as the sample mean, ratio, regression, and stratified estimators indispensable. This paper reviews the theoretical foundations of these estimators, emphasizing their key properties—unbiasedness, consistency, efficiency, sufficiency, and minimum mean square error (MMSE). Analytical expressions for bias and mean square error health, and social sciences are discussed, along with modern developments such as bootstrap and jackknife resampling, Bayesian estimation, and approaches for big data and adaptive sampling. The paper highlights the balance between unbiasedness and efficiency, illustrating how classical and contemporary estimators remain central to reliable statistical inference in increasingly complex survey contexts.

Muscle vibration alters both perceived limb position and velocity by increasing muscle spindle afferent firing rates. In particular the type Ia afferents are affected, which mainly encode muscle stretch velocity. Predictive frameworks of sensorimotor control, such as Active Inference and Optimal Feedback Control, suggest that velocity signals should inform position estimates. Such a function would predict that errors in perceived limb position and velocity should be correlated, but this prediction remains empirically underexplored. We hypothesised that an online evaluation of the integral of sensed velocity influences the perceived arm position during active movements. Using a virtual reality-based reaching task we investigated how vibration-biased proprioceptive feedback influences voluntary movement control and inference of arm position and movement. Our results suggest that muscle vibration biases perceived movement velocity, with downstream effects on perceived limb position and reflexive corrections of movement speed. We found that (i) antagonist vibration during active movement caused participants to overestimate their movement speed while also slowing down, (ii) movement speed and endpoint errors were correlated, with muscle vibration affecting both in congruent directions and (iii) adjustments in movement speed to muscle vibration are sufficiently fast to be reflexive. Together these findings support the hypothesis that proprioceptive velocity signals are integrated to augment inference of position, consistent with predictive frameworks of sensorimotor control. KEY POINTS: During movement without visual feedback, the central nervous system (CNS) has access to both position- and velocity-based proprioceptive signals, which are used to estimate limb state. Muscle vibration biases the perception of limb position, as seen in the classically observed pattern of biased endpoint errors, through the stimulation of primary (type Ia) muscle spindles, primarily a velocity sensor. We investigated how proprioceptive velocity signals affect position estimation during movement by applying muscle vibration while measuring perceived movement speed, actual movement speed and endpoint errors in a virtual reality (VR)-based reaching task. We show that errors in perceived limb position and velocity are correlated during active movements, consistent with predictive frameworks of sensorimotor control. These findings support the idea that the CNS maintains a self-consistent estimate of limb state across both position and velocity domains.

Multi-source precipitation fusion for hydrological models: Correction and metrics importance analysis

India is home to more than 525 million ruminants, which are major contributors to global warming via enteric methane (EntCH4) emissions. Various mitigation strategies exist to reduce EntCH4 emissions but accurate emission estimates are needed to establish the true potential of these strategies. Measuring EntCH4 emissions is expensive and unrealistic at such a large scale, so an urgent need exists for accurate EntCH4 prediction models. The present study evaluated the accuracy of various existing models and developed a new model to predict EntCH4 emissions from cattle in India. Six EntCH4 prediction models based on either DMI or gross energy intake (GEI) were identified as applicable to and suitable for the Indian context. Models based on DMI and GEI were derived from various works, including those of the Intergovernmental Panel on Climate Change and others (designated as IPCCDMI, IPCCGEI, RibeiroDMI, RibeiroGEI, PatraDMI, and PatraGEI). These were evaluated using 2 independent datasets characterizing 528 lactating (dairy) and 122 nonlactating (nondairy) cattle from 15 and 13 studies, respectively, under different management practices across 13 Indian states. Furthermore, the same datasets were combined to develop an empirical EntCH4 prediction model using a linear mixed-effects framework. The relative prediction error (RPE) and mean bias error (MBE) were used to evaluate model accuracy. A model's prediction was considered acceptable when RPE was <20%. None of the 6 models predicted EntCH4 for nondairy cattle with an RPE <20%. None of the 6 models predicted EntCH4 for both dairy and nondairy cattle with an RPE <20%. For dairy cattle, only the RibeiroDMI and PatraDMI models approached this threshold, producing RPE values of 22.6% and 22.9%, respectively. The linear mixed-effects model (Alam's model, described herein: EntCH4 [g/d per head] = 15.45 + 1.91 × DMI [kg/d], conditional R2 = 0.94), developed for both dairy and nondairy cattle, achieved a substantially lower RPE (9.48%) than any of the 6 previously tested models. Whereas the RibeiroDMI and PatraDMI models could acceptably predict EntCH4 emissions from dairy cattle in India, none of the evaluated models were suitable for nondairy cattle. Our linear mixed-effects model provides more accuracy than the latter 2 in estimating emissions for dairy cattle and also offers a suitable option for nondairy cattle in India.

A reduced-order model for predicting transient performance of air-source heat pumps

This study evaluated a purpose-trained markerless motion capture system for biomechanical analysis in elite pole vault competition. The aim was to determine whether a markerless approach could produce results comparable to manual digitising-the current standard in live competition settings-when operating under the practical constraint of a fixed four-camera setup. Data were collected from eight world-class pole vaulters during the 2024 World Athletics Indoor Championships. The final steps of the run-up through take-off were recorded at 100 Hz and analysed using both manual digitisation and the SIMI Nemo Markerless system to extract key biomechanical variables. Results showed strong overall agreement between methods for most spatial and centre of mass (CM) variables, with mean relative bias and random error of 0.3% and 3.8%, respectively. Step length differed by approximately 1 cm, and running step velocities showed root mean square error (RMSE) values between 0.02 and 0.05 m/s. CM height and horizontal velocity at pole plant showed RMSEs below 0.02 m and 0.1 m/s, respectively. At take-off, horizontal, vertical and absolute CM velocities all showed RMSE values of approximately 0.1 m/s. For these variables, intraclass correlation coefficients ranged from 0.898-1.000. Continuous waveform agreement was also strong, with Coefficient of Multiple Determination values exceeding 0.98 for vertical CM displacement, and above 0.90 for both CM velocity and most joint angle trajectories. In contrast, joint angles at take-off showed less agreement (RMSE 5°-10°), reflecting challenges in joint landmark identification in field conditions, and indicating that further refinement may be needed for complex movements. These findings suggest that, when supported by anatomically-informed pose estimation algorithms, a four-camera markerless setup is capable of capturing essential performance indicators in elite pole vault. The approach shows strong potential for scientific and applied use in real-world sport environments.

Accurate measurement of dietary intake remains a cornerstone challenge in optimizing the efficacy of nutritional interventions in human disease. Traditional self-reporting methods, although scalable and widely used, are prone to major bias and measurement error, thereby limiting their precision and clinical utility. In this Review, we highlight recent advances in technology-assisted food intake measurement, including image-based logging, wearable sensors and artificial intelligence (AI)-based dietary estimation, which may reduce reliance on recall and improve intake estimation. We review the emergence of non-invasive biological methodologies, such as metagenome-informed metaproteomics, in accurately enabling objective measurement of food intake and nutrient digestion and absorption in molecular resolution. We explore the possible interactions and effects of the gut microbiome in modulating such person-specific digestive and absorptive patterns and discuss challenges and prospects in the convergence of omics-based, measurement-based and AI-based dietary assessment tools into precision nutrition, in fulfilling its immense potential towards optimization of patient care.

Photon loss rates set an effective upper limit on the size of computations that can be run on current linear optical quantum devices. We present a family of techniques designed to mitigate the effects of photon loss on both output probabilities and expectation values derived from noisy linear optical circuits composed of an input of n photons, an m --mode interferometer, and m single photon detectors. Central to these techniques is the construction recycled probabilities . Recycled probabilities are constructed from output statistics affected by loss, and are designed to amplify the signal of the ideal (lossless) probabilities. Classical postprocessing techniques then take recycled probabilities as input and output a set of loss-mitigated probabilities, or expectation values. Our postprocessing methods result in biased estimators of the lossless probabilities. Nevertheless, we provide both analytical and numerical evidence that these methods can be applied, up to large sample sizes, to produce output probabilities with lower combined bias and statistical errors than the statistical errors of the output probabilities obtained from postselection. Therefore, these methods can outperform postselection - currently the standard method of coping with photon loss when sampling from discrete variable linear optical quantum circuits. In contrast, we provide evidence that the popular zero-noise extrapolation technique cannot improve on the performance of postselection for any photon loss rate.

Laboratory errors represent a critical yet underestimated threat to patient safety, with 26-30 % of reported errors adversely affecting patient care. Despite extensive scientific literature, no sustainable, open-access platform exists that allows laboratory professionals to rapidly access comprehensive information on error types, bias magnitudes, and clinical risks for specific analytes and platforms. To address this gap, the European Federation of Clinical Chemistry and Laboratory Medicine (EFLM) established the Committee on Laboratory Error Database (C-LED). C-LED aims to develop and maintain a continuously updated, evidence-based database covering laboratory errors, their impacts on test results, severity of potential harm, and supporting references across multiple analytical platforms and assay generations. C-LED employs a novel three-pillar information procurement strategy encompassing literature knowledge, manufacturer data, and supplementary sources, including incident reports and external quality assessment data. The literature procurement utilizes an innovative two-stage AI-driven approach: Stage 1 employs Claude Desktop with Model Context Protocol integration for systematic PubMed and Crossref screening, while Stage 2 combines Google NotebookLM and Claude Desktop for comprehensive full-text data extraction using standardized prompts and metadata structures. A case study examining haemolysis interference on cardiac troponin measurements demonstrates the approach's value, revealing platform-specific patterns ranging from minimal interference in contemporary high-sensitivity assays to severe bias in older platforms. The study identified that interference magnitude is concentration-dependent, with greater impacts at lower analyte baselines. By combining institutional governance, AI-driven efficiency, and multiple information sources, C-LED aim to establish a sustainable framework for database construction, ultimately enhancing diagnostic accuracy and patient safety globally.