ABSTRACT With intensifying hydroclimatic nonstationarity and human regulation, seasonal runoff peaks have become erratic, undermining conventional deterministic forecasts. A peak-sensitive hybrid framework is introduced by coupling time-varying filtering–based empirical mode decomposition (TVF-EMD) with deep learning. Monthly runoff is split into general and complex components. The deterministic backbone uses Bayesian optimization (BO)-tuned convolutional neural network (CNN)-bidirectional gated recurrent unit (BiGRU) with self-attention (SA), plus an observation-linked error correction (OLEC). Across four metrics it delivers superior point predictions, with Willmott’s index (WI) ranging from 0.9990 to 0.9997, Nash-Sutcliffe efficiency (NSE) from 0.9960 to 0.9988, low mean absolute error (MAE), and percent bias (PBIAS) between −1.1682% and 0.7844%. For the complex component, quantile regression (QR) with an error-sensitive focal loss (ESFL) produces calibrated yet sharper intervals. At 90% nominal coverage, prediction interval coverage probability (PICP) spans 98.6111−100% and prediction interval normalized average width (PINAW) spans 6.8663−23.9493%, with consistently lower Winkler scores than the BO-CNN-BiGRU-SA-QR baseline. Overall, the framework yields narrower, well-calibrated, peak-aware prediction intervals that support risk-informed water-resources management.

This paper investigates the estimation of the process capability index (PCI), C s , for the Generalized Inverse Lindley Distribution (GILD) under the Multiple Interval Type-I Censoring Scheme (MITICS). The main contribution of this study lies in extending existing estimation techniques to censored reliability data, providing a unified framework for evaluating process capability under complex censoring conditions. To estimate C s , three approaches are considered: maximum likelihood estimation, maximum product spacing, and Bayesian inference. The Bayesian estimates are developed using both the likelihood and product spacing functions under two loss functions–the squared error loss function and the linear exponential loss function–assuming gamma priors for the model parameters. In addition to point estimates, approximate confidence intervals are constructed using classical methods and compared with highest posterior density credible intervals obtained through the Bayesian approach. The optimal censoring plan is examined under different optimality criteria. A detailed Monte Carlo simulation study is carried out to evaluate the performance of the proposed estimators in terms of mean squared error, coverage probability, and average interval width. To illustrate the practical utility of the methods, two real-life data sets are analyzed: one related to breaking strengths of single carbon fibres and the other concerning the bond strengths of laser welds. The results demonstrate the effectiveness of the proposed techniques in assessing process capability under complex censoring schemes.

Desirability of outcome ranking (DOOR) is a paradigm for the design, monitoring, analysis, interpretation, and reporting of clinical trials based on patient-centric benefit-risk evaluation, developed to address limitations of existing approaches and advance clinical trial science. The first step in implementing DOOR is defining an ordinal DOOR outcome representing a global patient-centric response, a cumulative summary of the benefits and harms for an individual patient. This article aims to develop an analysis methodology for the setting where the DOOR outcome is a progressive time-varying state, and there is interest in event times and times that patients spend in more and less desirable states. We develop methods to estimate and make inferences about the temporal treatment effects. If the k-levels of the DOOR outcome are monotone, then k - 1 non-overlapping Kaplan-Meier survival curves can be estimated and plotted. The areas under the curves asymptotically follow a multivariate Gaussian distribution. We apply restricted mean survival time (RMST) concepts to the ordinal Kaplan-Meier curves and provide steps for estimating the covariance structure. Simulation studies demonstrate that the proposed methods perform well in practical settings. We generate censoring time under a uniform distribution and event times under a multi-state structure. The proposed estimators have small biases, the 95% confidence intervals have correct coverage probabilities, and the proposed tests accurately control the type I error rate under the null hypothesis. We illustrate the methods using data from Adaptive COVID-19 Treatment Trial (ACTT-1), a clinical trial that compared remdesivir vs placebo for the treatment of COVID-19 infection. Ordinal DOOR outcomes, which incorporate benefits and harms and represent an overall patient response, have recently been recommended by the Council for International Organizations of Medical Sciences (CIOMS) as a standard approach to benefit:risk analysis. Such endpoints recognize the cumulative nature of outcomes on patients, account for correlations between efficacy and safety, incorporate multivariate survival outcomes, offer generalizability to inform clinical practice, and recognize finer gradations of patient response and binary outcomes. Robust and interpretable analysis methodologies for ordinal outcomes are needed. Restricted mean survival time is a useful nonparametric approach for robust treatment effect estimation. We provide a framework for inference using multiple RMSTs to analyze DOOR and other ordinal outcomes using an interpretable time metric.

This paper proposes a representative point-based bootstrap (RP-bootstrap) to improve confidence interval estimation for the logistic distribution. The method replaces the traditional empirical distribution with a smoothed approximation constructed from statistically optimal representative points (RPs), leading to a more stable resampling distribution. We integrate the RP-bootstrap with the bootstrap-t, percentile, and BCa methods to construct intervals for the location and scale parameters. Its performance is compared to the classical nonparametric bootstrap via comprehensive Monte Carlo simulations and two real-data applications. The results show that the RP-bootstrap delivers noticeable improved finite-sample performance, particularly for small samples where standard bootstrapping often under-covers. It achieves recognizably higher coverage probabilities while maintaining shorter or comparable expected interval lengths. These improvements are strongest for the bootstrap-t interval and are consistent for both parameters, though more marked for the location. In conclusion, the RP-bootstrap is a computationally efficient and reliable alternative for logistic inference, offering enhanced accuracy, especially in small-sample scenarios. Purpose: Construction of confidence intervals under small sample size is frequently encountered in statistical inference, such as estimating some treatment effect in medical research with limited number of patients. Traditional nonparametric bootstrap methods often suffer from undercoverage in such settings. To address this limitation, we propose the RP-bootstrap—a resampling procedure that draws samples from an approximated distribution formed by representative points (RPs) of the logistic distribution. Methods: The RP-bootstrap is developed for constructing confidence intervals for the mean and variance of the logistic distribution. The algorithm generates weighted samples from the estimated RPs. The RP-bootstrap method is applied to construction of different types of confidence intervals (CIs) like the bootstrap-t, percentile, and {\rm BCa} CIs. Its performance and comparison with the traditional nonparametric bootstrap are evaluated through Monte Carlo simulation and real-data application. Results: Based on the Monte Carlo study under a set of small sample sizes, the RP-bootstrap achieves noticeable higher empirical coverage probability and competitive or shorter expected interval lengths compared with the nonparametric bootstrap. The improvements are much noticeable for small sample sizes like n<30 and for the bootstrap-t confidence intervals, where the nonparametric bootstrap frequently shows undercoverage of the true population parameter. Contribution: This study demonstrates that representative points provide a stable and efficient alternative to resampling methods from logistic models. The RP-bootstrap offers a practical method for reliable small-sample inference and yields confidence intervals with improved accuracy and reduced variability relative to the traditional nonparametric bootstrap method.

BackgroundNeoadjuvant chemoradiotherapy for locally advanced rectal cancer yields pathological complete response rates of only 10%–20%. Dose‐escalation strategies may improve outcomes, but optimal GTV‐to‐PGTV margins for CT‐guided radiotherapy with daily IGRT remain undefined.MethodsTwelve LARC patients undergoing CT‐guided daily IGRT with a simultaneous integrated boost were included. Daily diagnostic‐quality fan‐beam CT (FBCT) scans were acquired for IGRT. GTV and CTV were delineated on planning CT and all FBCTs. Target coverage margins were assessed by isotropically expanding the planning GTV until more than 95% of the voxels of the sequential GTVs were covered. A margin with a coverage probability threshold of 90% was defined as adequate. An independent validation cohort of 30 patients who underwent weekly FBCT‐guided image guidance was further analyzed. Overlap volumes between PGTVs and organs‐at‐risk (OARs; bladder and small bowel) were calculated to assess OAR sparing.ResultsAnalysis of 286 FBCT scans showed that a 6 mm isotropic GTV‐to‐PGTV margin achieved>95% coverage in>90% of fractions. Compared with 10 mm expansion, a 6 mm PGTV reduced the overlap volumes with the bladder and small bowel by 68.5% and 68.4%, respectively. A 6 mm isotropic expansion achieved>95% coverage in 91.3% of fractions in the validation cohort.ConclusionA 6 mm isotropic GTV to PGTV margin provides adequate target coverage for most middle‐ and lower‐rectal tumors while reducing OAR overlap. This finding could facilitate safer dose escalation while maintaining target coverage. However, larger margins may be necessary for smaller tumors or those located in the high rectum.

Abstract To address the limitations of existing stay-cable assessment systems—which rely on single indicators and fail to quantify long-term evolutionary trends and associated risks—this study proposes a dual-dimension assessment framework that integrates real-time condition evaluation with dynamic prediction. The framework establishes a complete technical closed loop encompassing the stable extraction of state features from measurement data, accurate prediction of long-term trends, rational quantification of predictive uncertainty, and ultimately, the generation of standardized maintenance decisions. It achieves coordinated evaluation between static equilibrium dead-load cable force (DLCF) and time-varying risk predicted cable force. The weights of the two dimensions are determined through logistic regression, and the resulting scores are mapped to code-based deduction values (DP) to guide graded maintenance decision-making. To support this framework, a multi-stage interference separation method is developed to robustly extract the DLCF, and a causally dilated attention-based bi‐directional long short‐term memory prediction model based on a causal dilated attention mechanism is constructed, incorporating a dynamic weighting strategy to enhance long-term prediction accuracy. Validation on a three-tower, low-pylon cable-stayed bridge demonstrates that the DLCF fluctuation is stably controlled within ±0.5% of the design value. The prediction model achieves an RMSE of 0.9 kN and an MAE of 0.5 kN, with a 95% confidence-level coverage probability (CP) of 97.65% and an average interval width of only 10.83 kN. Compared with classical time-series models, the RMSE is reduced by 35.7% and CP is improved by 2.5%. The composite assessment scores show a high level of consistency with maintenance records. The proposed systematic framework provides a generalizable measurement-analysis pathway for the service-safety evaluation of stay cables and other cable-supported structures.

In the context of comparing a treatment group to a control group, a mixture model for the observations from the treatment group allows for being able to make inferences about the existence of responders and non-responders in the treatment group. A generalised treatment effect for the model is represented by the probability that a treated patient is a responder and the magnitude of a shift in the control group distribution that models the responses from patients who respond to the treatment. Pseudolikelihood (PSL) and method of moment (MOM) estimators for the generalised treatment effect are derived, which account for the inclusion of covariates. Confidence intervals based on the asymptotic properties of the MOM estimator are also developed. Except when the overall treatment effect is small, simulation results demonstrate that the PSL estimator is preferred over the MOM estimator and that the confidence intervals have satisfactorily close to nominal coverage probabilities.

EEG microstate biomarkers for major depressive disorder: A comparative analysis using two independent datasets.

In this study, the maximum likelihood estimators (MLEs) and Bayes estimators for the shape and scale parameters of Inverse Exponential Power (IEP) distribution are derived. As closed-form solutions for the Bayes estimators are not available, approximate estimators are obtained through Lindley’s and Tierney–Kadane’s approximation methods, along with the Markov Chain Monte Carlo (MCMC) method, under the squared-error loss (SEL) function. Also, the approximate Bayes estimates are evaluated against the maximum likelihood estimates based on mean square error (MSE) and bias values using Monte Carlo simulation. In addition, the coverage probabilities of the parametric bootstrap estimates are computed. Finally, real data sets belonging to the COVID-19 Pandemic Case Fatality Rate across World Health Organization (WHO) and Organization fo Economic Co-Operation and Development (OECD) regions data is investigated as an important indicator to achieve United Nations’ Sustainable Development Goal 3 (SDG 3) are employed to display the emprical results associated with both non-bayesian and bayesian estimations of the IEP distribution presented. By offering improved estimation techniques for flexible health indicator distributions, the results contribute to the broader effort of enhancing statistical tools used in global health analytics—particularly in areas such as survival modeling, biomedical reliability, and chronic disease monitoring aligned with SDG 3.Supplementary InformationThe online version contains supplementary material available at 10.1038/s41598-025-27264-7.

Objective.Urodynamic studies (UDSs) are vital for evaluating bladder function but require invasive catheterization to measure detrusor pressure (Pdet). Our purpose is to report on technical developments toward a noninvasive and accurate estimation ofPdetthrough ultrasound (US), a technique we refer to as US-UDS.Approach.The proposed US-UDS technique produces an estimate ofPdetby using US to induce and track elastic waves that propagate within and around the bladder wall. Key components of the US-UDS processing include: an US sequence with dense time sampling; an empirical correction accounting for deviation from a Lamb wave dispersion model; and constrained optimization forPdetestimation. US-UDS analysis for estimation ofPdetwas applied to 533 distinct data points collected from 3 human subjects undergoing concurrent UDS. Concordance analysis was used to evaluate agreement between US-UDS and UDS measures ofPdetwhile response operator characteristic analysis was used to evaluate US-UDS for detection of detrusor activity (parameterized as UDSPdet>15 cmH2O).Main results.US-UDS was able estimatePdetwithin 10 cmH2O and 5 cmH2O of UDS measurements with coverage probabilities of 95.7% and 68.8% respectively. Furthermore, US-UDS was able to detect UDSPdetmeasurements greater than 15 cmH2O with a sensitivities and specificities of 0.99 and 0.83 using a fixed threshold (>15 cmH2O for US-UDS) and 1.0 and 0.93 for an optimized threshold (> 11.08 cmH2O for US-UDS).Significance.Our results show that the technical developments of US-UDS put forth in this work are able produce accurate and useful estimates ofPdetas compared to traditional UDS. Although additional research with a large number of subjects is needed to fully characterize the clinical utility of US-UDS, the developments and results from this work demonstrate that clinically useful non-invasive measurements ofPdetare feasible from US measurements.

Solid oxide fuel cells (SOFCs) are becoming increasingly important due to their high electrical efficiency, the flexible choice of fuels and relatively low emissions of pollutants. However, the increasing demand for electrochemical devices, coupled with ongoing challenges like significant degradation mechanisms, necessitates further technological advancements to fully realize the potential of fuel cells. Since the 3D electrode morphology—strongly influenced by the manufacturing process—plays a crucial role with regard to the electrochemical performance, a deeper understanding of structural changes induced by process modifications or degradation phenomena is essential.Therefore, we investigate the influence of the annealing time, operating temperature and manufacturing process on the 3D morphology of SOFC anodes using 3D image data obtained by focused-ion beam scanning electron microscopy [1]. The resulting image data is then segmented into gadolinium-doped ceria, nickel and pore space. To quantitatively analyze structural differences between these SOFC anodes, various geometrical descriptors such as the specific surface area, the specific surface area of phase interfaces as well as the specific length of the triple phase boundary, where the chemical reaction takes place, are used. Additionally, the geodesic tortuosity, which quantifies the length of shortest transport paths, the constrictivity (a measure for the strength of bottleneck effects), the two-point coverage probability function, the spherical contact distribution and the chord length distribution are computed. This enables an extensive statistical analysis of the complex 3D morphology. Moreover, these descriptors are not only computed globally, but also locally to quantify the heterogeneity of the anode structure.[1] Weber, S., Prifling, B., Juckel, M., Liu, Y., Wieler, M., Schneider, D., Nestler, B. & Schmidt, V. (2024). Comparing the 3D morphology of solid-oxide fuel cell anodes for different manufacturing processes, operating times, and operating temperatures. arXiv preprint arXiv:2411.15259.

To tackle the multi-step error amplification, over-wide uncertainty intervals, and insufficient long-range dependency modeling that plague complex dynamic time series, we propose a lightweight localglobal coupled forecasting framework. The approach employs sliding residual convolution to extract fine-grained local variations, combines sparse attention to capture long-cycle global correlations, and introduces quantile regression to output tightly coupled prediction intervals, achieving joint optimization of point forecasts and uncertainties. Evaluated on 17 520 real-world electricity-load samples, the proposed model reduces 24-, 48-, and 96-step RMSE to 0.862, 0.921, and 1.012 (10 kWh), respectivelyan average 14.7 % improvement over Informerwhile its 95 % prediction-interval coverage probability (PICP) reaches 94.7 % and interval width (PINAW) is only 1.9810, significantly outperforming LSTM-Q and DeepAR. Results demonstrate that the framework achieves a better balance among accuracy, robustness, and reliability, and can be directly embedded in real-time scheduling and risk-decision systems for energy, transportation, and finance.

Major depressive disorder (MDD) is closely linked to suicidal ideation (SI), which has been associated with cognitive dysfunction and abnormal brain network connectivity. However, the relationships among SI, cognitive functioning, and large-scale brain dynamics in MDD remain unclear. Resting-state EEG microstate analysis provides a novel method for exploring these associations by capturing transient global brain activity patterns. Eighty-eight drug-naïve, first-episode MDD patients were divided into SI (n = 54) and noSI (n = 34) groups, with 34 age- and gender-matched healthy controls. SI was assessed using the Beck Scale for Suicide Ideation (BSSI), and cognition was evaluated with the MATRICS Consensus Cognitive Battery (MCCB). Resting-state EEG was recorded, and microstate parameters were analyzed. Group differences were tested with ANOVA and ANCOVA, correlations examined with Pearson and partial analyses, and multiple comparisons controlled by FDR correction. The noSI group showed greater microstate D coverage (p = 0.037) and higher transition probability from microstate A to D (p = 0.011) than the SI group; these effects remained after controlling for depressive and anxiety symptoms and agitation and FDR correction. In the SI group, both measures correlated positively with SDMT scores, and SI severity was negatively related to microstate C coverage and transition probability from microstate B to C; however, none of these correlations survived adjustment. MDD patients with SI show alterations in EEG microstate dynamics that are associated with reduced processing speed. These findings indicate that large-scale brain network instability may contribute to neurocognitive dysfunction related to suicidality. EEG microstates may provide preliminary insights into its neural basis and represent exploratory biomarkers that require replication and longitudinal validation. Not applicable. This study is not a clinical trial. It is an observational study designed to investigate the relationship between EEG microstates and cognitive function in depressed patients with and without suicidal ideation. Therefore, no clinical trial registration number is applicable to this research.

This study addresses the estimation of the common mean for the zero-inflated inverse Gaussian (ZIIG) distributions, a problem not previously explored. The performance of four interval estimation approaches was evaluated: the generalized confidence interval (GCI), parametric bootstrap, Bayesian, and highest posterior density (HPD). Simulation studies under varying sample sizes, zero-inflation probabilities, mean values, and shape parameters revealed notable differences in coverage probability (CP) and average length (AL). For small samples, the GCI and parametric bootstrap approaches often under-covered, particularly in highly skewed or heavily zero-inflated cases. In contrast, Bayesian and HPD intervals generally maintained coverage closer to the nominal 0.95 level, albeit with longer intervals. As sample size increased, all methods approached nominal coverage and produced shorter intervals, improving precision. Overall, the Bayesian and HPD approaches demonstrated strong robustness across conditions, with HPD intervals frequently achieving accurate coverage with shorter lengths. Finally, the proposed approaches were applied to real-world data on road accident fatalities in Thailand.

Scientific, accurate, and interpretable carbon price forecasts provide critical support for addressing climate change, achieving low-carbon goals, and informing policy-making and corporate decision-making in energy and environmental markets. However, the existing studies mainly focus on deterministic forecasting, with obvious limitations in data feature diversity, model interpretability, and uncertainty quantification. To fill these gaps, this study constructs an interpretable hybrid system for carbon market price prediction by combining feature screening algorithms, deep learning models, and interpretable explanatory analysis methods. Specifically, this study first screens important variables from twenty-one multi-source structured and unstructured influencing factor datasets on five dimensions affecting carbon price using the Boruta algorithm. Immediately after that, this study proposes a hybrid architecture of bidirectional temporal convolutional network and Informer models, where a bidirectional temporal convolutional network is used to extract local spatio-temporal dependent features, while Informer captures long sequences of global features through the connectivity mechanism, thus realizing staged feature extraction. Then, to improve the interpretability of the model and quantify the uncertainty, this study introduces Shapley additive explanations to analyze the feature contribution in the prediction process, and the Monte Carlo dropout method is used to achieve interval prediction. Finally, the empirical results in China’s Guangdong and Shanghai carbon markets show that the proposed model significantly outperforms benchmark models, and the coverage probability of the obtained prediction intervals significantly outperforms the confidence level. The Shapley additive explanation analysis reveals regional heterogeneity drivers. In addition, the proposed model is also intensively validated in the European carbon market and the U.S. natural gas market, which also demonstrate an excellent prediction performance, indicating that the model has good robustness and applicability.

The Poisson distribution is commonly used when events are assumed to be independent and occur at a consistent rate. This may not be generally applicable, and the Poisson distribution is not appropriate in situations where the underlying rate of occurrence displays variability. A mixed Poisson distribution such as the Poisson-Rani distribution permits the rate parameter to be random instead of constant. Bootstrap-based confidence intervals (CIs) were developed for the Poisson-Rani distribution parameter in this study. The percentile bootstrap (PB), basic bootstrap (BB), and bias-corrected and accelerated (BCa) bootstrap methods were compared for empirical coverage probabilities and expected lengths by the Monte Carlo simulation using the RStudio program with sample sizes of 10, 30, 50, 100, 500, and 1,000. The parameter values were set at 0.1, 0.3, 0.5, 0.8, 1, 1.5, and 2 with 1,000 replications. The simulation results suggested that the bootstrap-based CIs required improvement to attain the nominal confidence level for small sample sizes. No significant differences were detected in the performances of bootstrap-based CIs when evaluating large sample sizes, with the BCa bootstrap CI exhibiting superior performance compared to the others. The application of bootstrap-based CIs to meteorological data yielded comparable results to the simulation study.

BackgroundBehavioral and psychological symptoms of dementia (BPSD) are common in Alzheimer's disease (AD), yet their mechanisms remain unclear.ObjectiveWe aim to explore the possible neurophysiological mechanisms of BPSD using high temporal resolution electroencephalography (EEG) microstate technology, laying the foundation for clinical evaluation and subsequent treatment.MethodsWe enrolled 52 AD patients (25 with BPSD, 27 without) and 29 age- and gender-matched healthy controls (HC). All participants underwent various neuropsychological assessments and resting-state EEG recordings. Resting-state EEG data were analyzed employing microstate analysis techniques, with a focus on four key microstate parameters: duration, occurrence, coverage, and transition probability. Inter-group comparisons were performed using post-hoc tests, with statistical significance determined through False Discovery Rate (FDR) correction. Furthermore, the correlations between the indicators and neuropsychological assessment scores were analyzed.ResultsCompared to the HC and non-BPSD groups, the BPSD group showed an increase in the transition rate from microstate A to microstate C. Compared to the HC group, the BPSD group showed an extension in the duration of microstate A and a decrease in the frequency of microstate D. Compared to the HC group, the non-BPSD group showed prolonged durations (A, B, mean) and reduced occurrences (C, D, mean).The partial correlation analysis with years of education as a covariate showed that in the BPSD group, the duration of microstate A was correlated with the severity of the Neuropsychiatric Inventory (NPI) and the Hamilton Anxiety Scale (HAMA).ConclusionsAD with and without BPSD exhibits different altered brain dynamics.

In the cooperative search for dynamic targets by multiple UAVs, target uncertainty and system complexity pose significant challenges to cooperative decision-making. Multi-agent reinforcement learning (MARL) technology can be used for cooperative policy optimization, but it suffers from convergence difficulties and low policy quality in reward-sparse environments such as dynamic target search. To address this issue, this paper proposes a Multi-Potential-Field Fusion Reward Shaping MAPPO (MPRS-MAPPO) algorithm. First, three potential field functions are constructed for reward shaping: probability edge potential field, maximum probability potential field, and coverage probability sum potential field. Subsequently, an adaptive fusion weight mechanism is proposed to adjust fusion weights based on the correlation between potential field values and advantage values. Furthermore, a warm-up phase is introduced to improve training stability. Extensive experiments, including multi-scale and physical tests, demonstrate that MPRS-MAPPO significantly improves convergence speed, detection rate, and stability compared with MAPPO, MASAC, QMIX, and Scanline. Detection rates increased by 7.87–29.76%, and training uncertainty decreased by 7.43–56.36%, validating the algorithm’s robustness, scalability, and real-world applicability.

Abstract Introduction and Research Gap This study presents a comprehensive framework for predicting volumetric water content (VWC) to mitigate shallow, rainfall-induced landslides, bridging existing gaps in the literature. Methodology The framework synergistically integrates the empirical strengths of deep learning (DL) with the physical dynamics of the VWC subsurface behavior. Statistical, shallow machine learning (ML), and DL models were investigated with optimization techniques and sensitivity analyses to establish benchmarks for comparison and derive optimal predictions. DL and probability theory enable both point and interval predictions. Findings Validation on the Pa Mei landslide demonstrates strong performance with mean absolute errors (MAE) ranging from 0.35% to 1.22% and Predicted Interval Coverage Probabilities (PICP) from 0.86 to 0.91. Predicted VWC deviations were propagated into Factor of Safety (FOS) calculations, yielding robust performance metrics with R 2 and PICP of 0.89 and 0.85, respectively. Transferability is demonstrated at the Tung Chung landslide, where MAE ranges from 0.36% to 1.25% and PICP from 0.86 to 0.95. Significance This framework demonstrates improved accuracy and introduces a practical data-sharing mechanism to address monitoring challenges such as power consumption and data loss, offering a robust tool for hazard mitigation and decision support.

This study implements Bayesian along with non-Bayesian approaches to estimate the parameters of the three-parameter quadratic hazard rate distribution using hybrid Type-II censoring. The model expands upon linear hazard rate, exponential, and Rayleigh distributions. In the non-Bayesian framework, point estimates and survival and hazard functions are calculated using maximum likelihood estimation (MLE). Asymptotic confidence intervals are derived, with a focus on the delta method. By applying independent normal and gamma priors, Bayesian inference produces point estimates and credible intervals using different symmetric and asymmetric loss functions. The analytical intractability of posterior distributions makes Markov chain Monte Carlo (MCMC) methods necessary for sampling purposes. The evaluation of point and interval estimates depends on root mean squared error (RMSE) in combination with mean relative absolute bias (MRAB), average confidence interval length (AL), and coverage probability (CP). The performance evaluation through different sample sizes and censoring schemes is conducted by simulation studies, while real-world data from COVID-19 mortality demonstrates the practical implementation of methods. Graphical and numerical analyses confirm the existence and uniqueness of the estimates. Results indicate that Bayesian methods deliver superior accuracy and more robust estimates than their non-Bayesian counterparts for survival analysis purposes in clinical and medical research.