Reliable Prediction Research Articles

Construction of brain age models based on structural and white matter information.

Brain aging is an inevitable process in adulthood, yet there is a lack of objective measures to accurately assess its extent. This study aims to develop brain age prediction model using magnetic resonance imaging (MRI), which includes structural information of gray matter and integrity information of white matter microstructure. Multiparameter MRI was performed on two population cohorts. We collected structural MRI data from T1- and T2-sequences, including gray matter volume, surface area, and thickness in different areas. For diffusion tensor imaging (DTI), we derived four white matter parameters: fractional anisotropy, mean diffusivity, axial diffusivity, and radial diffusivity. To achieve reliable brain age prediction based on structure and white matter integrity, we employed LASSO regression. We successfully constructed a brain age prediction model based on multiparameter brain MRI (Mean absolute error of 3.87). Using structural and diffusion metrics, we identified and visualized which brain areas were notably involved in brain aging. Simultaneously, we discovered that lateralization during brain aging is a significant factor in brain aging models. We have successfully developed a brain age estimation model utilizing white matter and gray matter metrics, which exhibits minimal errors and is suitable for adults.

Prediction of energy consumption in four sectors using support vector regression optimized with genetic algorithm.

Effectively managing and optimizing energy resources to accommodate population growth while minimizing carbon emissions has become increasingly intricate. A proficient approach to this dilemma is accurately predicting energy usage and emissions across diverse sectors. This paper unveils a genetic algorithm (GA)-optimized support vector regression (SVR) model designed to (i) predict electricity generation, (ii) predict energy consumption in four primary sectors-residential, industrial, commercial, and agricultural, and (iii) estimate sector-specific carbon emissions. The proposed model's efficacy is assessed by calculating the R2 value, mean absolute error (MAE), root mean squared error (RMSE), and residual plot. The model achieved high accuracy in predicting energy generation, with an MAE of 1.18%, and yielded reliable sectoral consumption predictions, reflected in MAE values of 1.22% (residential), 4.98% (industrial), 4.40% (commercial), and 4.04% (agricultural). The residuals exhibited homoscedasticity, and the R2 value approached one. The model predicts that by 2027, the residential sector will consume 55748.66GWh of energy, the commercial sector 14892.49GWh, the industrial sector 32642.35GWh, and the agricultural sector 2288.37GWh. It has been predicted that by 2027, these four sectors will release 75437.96-billion-gram carbon equivalents.

Reliability prediction of power electronic converters based on surrogate models

Reliability prediction of power electronic converters based on surrogate models

Physiologically-based pharmacokinetic modeling of enantioselective hydroxychloroquine kinetics and impact of genetic polymorphisms

Abstract Hydroxychloroquine (HCQ) is a chiral drug used to treat malaria and inflammatory diseases, available as a racemic mixture of R-and S-HCQ. This work aimed to build physiologically-based pharmacokinetic (PBPK) models to predict the pharmacokinetics (PK) of R-and S-HCQ and assess the impact of major genetic polymorphisms on PK. Whole-body PBPK models accounting for first-order absorption, Rodgers and Rowland distribution method, and enzyme kinetics data were built for R-and S-HCQ. The models were verified by comparing predicted PK parameters with observed ones, with a mean error within the acceptable range (0.5-2 fold). Simulations covered CYP2D6 normal metabolizer (NM), poor metabolizer (PM), and ultra-rapid metabolizer (UM) phenotypes, as well as CYP2C8 NM and PM phenotypes. The results revealed a 1.1-fold increase in area under the curve blood concentration versus time (AUC) for CYP2D6 PM individuals and a 0.9-fold reduction for UM individuals compared to NM individuals. In addition, simulations with CYP2D6 and CYP2C8 PM phenotype individuals combined with the CYP3A4 inhibitor clarithromycin showed increased AUC for R-and S-HCQ of 2.34 and 2.68, respectively. These PBPK models offer reliable predictions for R-and S-HCQ enantioselective kinetics and shed light on previously unexplored scenarios.

Sequence-variable attention temporal convolutional network for volcanic lithology identification based on well logs

A Sequence-Variable Attention Temporal Convolutional Network (SVA-TCN) is proposed for lithology classification based on well log data. This study aims to address the issue that native TCN pays insufficient attention to crucial logging variables and sequence structural features in well log tasks. A novel Sequence-Variable Attention Mechanism module is designed to effectively extract features by adding the bidirectional attention mechanism, which comprises the sequence attention mechanism and the variable attention mechanism. The sequence attention mechanism, acting along the depth dimension, focuses on modeling the multiscale sequence features. The variable attention mechanism, working along the logging variable dimension, contributes to learning the importance of different logging variables and exploring the internal correlations among them. To verify the validity of the model, experiments are conducted in the volcanic reservoir of the Yingcheng Formation in the Xujiaweizi Depression of the Songliao Basin. Seven logging variables sensitive to lithological changes are selected, including compressional wave slowness, density, and compensated neutron logging, etc., to construct a lithology identification model using SVA-TCN. Compared with machine learning and deep learning methods, the SVA-TCN demonstrates a remarkable accuracy of 99.00%, surpassing the accuracy of the comparison methods by 0.37–17.69%. The findings of this study indicate that the SVA-TCN model is well-suited for lithology identification in volcanic reservoirs. It provides more reliable prediction results and exhibits good stability and generalization, offering a new avenue to volcanic lithology identification.

Designing a Revenue Forecast Scheme for Analog Ensemble Method Applied Solar PV Forecasting in South Korea

Power companies have found that solar irradiance forecasting is a reliable method for anticipating and preparing for the intermittent nature of renewable energy sources (RESs). However, when the percentage of RESs in the energy mix rises, there is a negative correlation between forecasting accuracy and process error, which could impact not only the grid but also the whole RES’s economic sustainability. In order to tackle this problem, this paper examines the economic implications of employing the Analog Ensemble (AnEn) forecasting model within Korea’s solar energy sector. The levelized cost of energy (LCOE) model and the revenue prediction model are applied to assess the efficacy of the AnEn. The proposed revenue scheme scheme was initiated on a 1 MW photovoltaic (PV) ground‐type power plant deployed in the mainland of South Korea. Various methods have been examined to assess the financial benefit of utilizing the AnEn, such as forecast verification and LCOE and revenue forecast scheme. Key findings reveal that the AnEn model consistently outperforms traditional models in terms of accuracy, particularly during critical solar peak months. This improved accuracy translates into more reliable predictions of solar irradiance, essential for minimizing overestimations and supporting realistic financial planning. Furthermore, the paper explores the economic incentives implemented by the Korean government to encourage precision in energy forecasting, including a structured incentive scheme tied to forecast accuracy.

Research on reliability prediction method of fuel regulator based on failure rate model

Research on reliability prediction method of fuel regulator based on failure rate model

Linear regression model of structured column packing form factor

Semi-empirical correlations are commonly used to predict packed column pressure drop. As a downside, these models rely on packing-specific parameters that are obtained through laborious experiments.This work aims to eliminate packing-specific parameters for modeling pressure drop of structured packings. A multiple linear regression model is used to predict the packing form factor directly from packing properties. The form factor is then used with the model of Maćkowiak (2010) to calculate pressure drop. This tandem model approach therefore allows for pressure drop predictions directly from packing properties.The regression model includes model features to describe packing properties, such as packing geometry, perforations, packing material or packing surface. Model coefficients are determined using experimental pressure drop results of 28 different structured column packings manufactured from metal, plastic or ceramic. The fitted regression model yields very good (mean absolute error 1.86%) approximation of packing form factor. Model feature importance analysis suggests packing porosity, packing angle and packing perforations exert the greatest influence on the form factor.Leave-one-out cross-validation is performed to evaluate predictive model performance beyond training data. Results indicate the model’s reliable predictions of the form factor (mean absolute error 3.1%) and strong agreement between pressure drop predictions and experimental data (mean absolute error 23.4%).

Prediction accuracy of discrete choice experiments in health-related research: a systematic review and meta-analysis

Discrete choice experiments (DCEs) are increasingly used to inform the design of health products and services. It is essential to understand the extent to which DCEs provide reliable predictions outside of experimental settings in real-world decision-making situations. We aimed to compare the prediction accuracy of stated preferences with real-world choices, as modelled from DCE data. We searched six databases for health-related studies that used DCE to assess external validity and reported on predicted versus real-world choices, up to July 2024. A generalised linear mixed model was used for a meta-analysis to jointly pool the sensitivity and specificity. Heterogeneity was assessed using the I 2 statistic, and sources of heterogeneity using meta-regression. This study is registered with PROSPERO (CRD42023451545). We identified 14 relevant studies, of which 10 were included in the meta-analysis. Most studies were conducted in high-income countries (11/14, 79%) from the European region (9/14, 64%) and analysed using mixed logit models (5/14, 36%). Pooled sensitivity and specificity estimates were 89% (95% CI:77-95, I 2=97%) and 52% (95% CI:32-72, I 2=95%), respectively. The area under the SROC curve (AUC) was 0.81 (95% CI:0.77-0.84). Our meta-regression found that DCEs for prevention-related choices had higher sensitivity than treatment-related choices. DCEs conducted under clinical settings and analysed using the heteroskedastic multinomial logit model, incorporating systematic preference heterogeneity and random opt-out utility, had higher specificity than non-clinical settings and alternative models. DCEs are valuable for capturing health-related preferences and possess reasonable external validity to predict health-related behaviours, particularly for opt-in choices. Contextual factors (e.g., type of intervention, study setting, analysis method) influenced the predictive accuracy. JJO is supported by an Australian National Health and Medical Research Council Emerging Leadership Investigator Grant (GNT1193955). EBG is supported by the Dutch Research Council (NWO-Talent-Scheme-Vidi-Grant No, 09150171910002). YZ is supported by an Australian Government Research Training Program (RTP) scholarship.

Satellite-Driven Hydrological Modelling in Ungauged Moroccan bassins: Case study Ouzoud river

Predicting streamflow is vital for flash-flood early warning systems and managing water resources under climate change. However, limited streamflow observations restrict advanced prediction techniques to gauged basins, leaving most of the world's ungauged basins at a disadvantage. Developing reliable prediction methods for ungauged basins (PUB) is therefore essential. Over the past two decades, satellite-driven products, such as ERA5, have become crucial for enhancing precipitation and meteorological measurements, especially in complex terrains and changing climatic conditions. This study focuses on Morocco's arid and semi-arid regions, where water resource management is critical for agriculture, urbanization, and economic stability. Using the ERA5 dataset, which provides high-resolution atmospheric information, the study evaluates satellite-derived precipitation against ground measurements from the Sgatt station at Bernat River on daily timescales. Various statistical metrics assess ERA5's performance in representing daily precipitation and its integration into the GR4J-CemaNeige model for flow simulation. Results highlight the potential of ERA5 in improving rainfall representation and hydrological modelling in ungauged basins, verifying its effectiveness in simulating rainfall events compared to ground-based observations. This work underscores the importance of satellite-driven products for enhancing hydrological predictions and supporting water management in data-scarce regions.

MACHINE-LEARNING-BASED ASYMPTOTIC HOMOGENIZATION AND LOCALIZATION OF SPATIALLY VARYING MULTISCALE CONFIGURATIONS MADE OF MATERIALS WITH NONLINEAR ELASTIC STRESS-STRAIN RELATIONSHIPS

This article aims to propose a general method in support of efficient and reliable predictions of both the global and local behaviors of spatially varying multiscale configurations (SVMSCs) made of materials bearing nonlinear history-independent stress-strain relationships. The framework is developed based on a complementary approach that integrates asymptotic analysis with machine learning (ML). The use of asymptotic analysis is to identify the homogenized constitutive relationship and the implicit relationships that link the local quantities of interest, say, the site where the maximum von Mises stress (MVMS) lies, with other on-site mean-field quantities. As for the implementation of the proposed asymptotic formulation, the aforementioned relationships of interest are represented by neural networks (NNs) using training data generated following a guideline resulting from asymptotic analysis. With the trained NNs, the desired local behaviors can be quickly accessed at a homogenized level without explicitly resolving the microstructural configurations. The efficiency and accuracy of the proposed scheme are further demonstrated with numerical examples, and it is shown that even for fairly complex multiscale configurations, the predicting error can be maintained at a satisfactory level. Implications from the present study to speed up classical computational homogenization (CH) schemes are also discussed.

Modeling of heteroaggregation driven buoyant microplastic settling: Interaction with multiple clay particles.

The ecological risk of microplastics (MPs) has received widespread attention, but understanding ecological risk starts with understanding environmental migration. Heteroaggregation is an important process that affects the vertical migration of MPs, and the mathematical model is a common tool used to project the migration behavior of MPs. However, the mathematical model based on the aggregation of MPs with one clay particle is not applicable to simulate the migration behavior of buoyant microplastic (BMP). Hence, this study developed a model for heteroaggregation of one BMP with multiple clay particles based on the Population Balance Equation, and the main factors affecting the sedimentation of BMP are clarified through parameter sensitivity analysis and scenario simulation. The results show that neglecting the interaction of one BMP with multiple clay particles in the mathematical model can underestimate the predicted settling concentration of BMP, especially in aqueous environments with higher clay concentration and salinity. The settling process of BMP is controlled by the heteroaggregation rate, which is sensitive to environmental conditions and insensitive to particle properties. This study emphasizes the importance of considering the interaction of one BMP with multiple clay particles in the future mathematical model, which will provide a more reliable prediction of the migration of MPs in aquatic environments.

Failure physics modeling and reliability prediction of electromagnetic relay based on permanent magnet degradation

Failure physics modeling and reliability prediction of electromagnetic relay based on permanent magnet degradation

Explainable AI for permeate flux prediction in forward osmosis: SHAP interpretability and theoretical validation for enhanced predictive reliability

Explainable AI for permeate flux prediction in forward osmosis: SHAP interpretability and theoretical validation for enhanced predictive reliability

Multi-level Δ-learning for predicting the radiative decay rate constant of phosphorescent platinum(ii) complexes

The multi-level Δ-learning method significantly advances kr prediction by bridging TDDFT results with experimental values. This approach improves prediction reliability, making it essential for accurate material evaluation.

Association Kinetics for Perfluorinated n-Alkyl Radicals.

Radical-radical reaction channels are important in the pyrolysis and oxidation chemistry of perfluoroalkyl substances (PFAS). In particular, unimolecular dissociation reactions within unbranched n-perfluoroalkyl chains, and their corresponding reverse barrierless association reactions, are expected to be significant contributors to the gas-phase thermal decomposition of families of species such as perfluorinated carboxylic acids and perfluorinated sulfonic acids. Unfortunately, experimental data for these reactions are scarce and uncertain. Furthermore, obtaining reliable theoretical predictions for such reactions is a laborious and computationally intensive task. In this work, the chemical kinetics of the various association/decomposition reactions producing/decomposing the C2-C4 series of unbranched n-perfluoroalkanes (C2F6, C3F8, and C4F10) are examined using state-of-the-art ab initio transition-state-theory-based master-equation calculations. The variable-reaction-coordinate transition-state theory (VRC-TST) formalism is employed in computing the microcanonical and canonical rates for the association reactions. Reaction thermochemistry is obtained via composite quantum chemistry calculations and the laddering of error-canceling reaction schemes via a connectivity-based hierarchy approach employing ANL1/ANL0-style reference energies. Lennard-Jones collision model parameters for the considered systems were estimated by a direct dynamics approach, and collisional energy transfer parameters were obtained from analogies to systems of similar size and heavy-atom connectivity. A one-dimensional master equation approach was used to convert the microcanonical rate coefficients from the VRC-TST analysis into temperature- and pressure-dependent rate constants for the association reactions and the reverse dissociation reactions. The data are reported in standardized formats for usage in comprehensive chemical kinetic models for PFAS thermal destruction.

Correlation and Prediction Analysis of Cyanobacteria with Water Environment Data

Objectives : A study was conducted to implement a predictive for cyanobacteria occurrence, a direct indicator of algae presence. Water quality, river environment, and meteorological data were collected and analyzed at six locations along the Nakdong River in Gyeongsangbuk-do and Daegu City. The primary objective was to establish correlations between various environmental factors and validate the utility of predicting algae(cyanobacteria) occurrences.Methods : Cyanobacteria was designated as the dependent variable for correlation analysis. Water quality parameters associated with algal growth, including chlorophyll-a, water temperature, pH, dissolved oxygen(DO), total nitrogen(TN), and total phosphorus(TP), were utilized as key indicators. Additionally, meteorological data such as air temperature, humidity, cloud cover, precipitation, and solar radiation, as well as flow rate and flow quantity, which are indicators of water body stability, were utilized. The collected data were verified and corrected for accuracy before analyzing correlations between cyanobacteria occurrence and the independent variables. Furthermore, cyanobacteria occurrence was predicted using a random forest algorithm, with a linear regression model serving as a baseline for comparison.Results and Discussion : The data presented a normal distribution. Correlation analysis indicated that previous cyanobacteria occurrences had a significant influence on current occurrences. Water temperature showed a positive correlation with cyanobacteria, while DO exhibited a negative correlation. Flow quantity and flow rate were inversely correlated with cyanobacteria cell density. In terms of meteorological data, air pressure negatively correlated with cyanobacteria occurrence, while air temperature showed a positive relationship. Substantial precipitation significantly reduced cyanobacteria concentrations; however, this effect was less pronounced during periods of low rainfall. The linear regression model, with an R2 value of 0.734, demonstrated considerable explanatory power, indicating its effectiveness in accounting for the variability of the dependent variable.Conclusion : The random forest, built on the linear regression model, successfully captured the overall trend of cyanobacteria occurrence when compared with field measurements. The model consistently maintained predictive performance by effectively recognizing complex data patterns, suggesting its potential for reliable prediction of cyanobacteria occurrence trends in aquatic ecosystems.

Uncertainty quantification and propagation in atomistic machine learning

Abstract Machine learning (ML) offers promising new approaches to tackle complex problems and has been increasingly adopted in chemical and materials sciences. In general, ML models employ generic mathematical functions and attempt to learn essential physics and chemistry from large amounts of data. The reliability of predictions, however, is often not guaranteed, particularly for out-of-distribution data, due to the limited physical or chemical principles in the functional form. Therefore, it is critical to quantify the uncertainty in ML predictions and understand its propagation to downstream chemical and materials applications. This review examines existing uncertainty quantification (UQ) and uncertainty propagation (UP) methods for atomistic ML under the framework of probabilistic modeling. We first categorize the UQ methods and explain the similarities and differences among them. Following this, performance metrics for evaluating their accuracy, precision, calibration, and efficiency are presented, along with techniques for recalibration. These metrics are then applied to survey existing UQ benchmark studies that use molecular and materials datasets. Furthermore, we discuss UP methods to propagate uncertainty in widely used materials and chemical simulation techniques, such as molecular dynamics and microkinetic modeling. We conclude with remarks on the challenges and opportunities of UQ and UP in atomistic ML.

Artificial Intelligence in Optimization of Cloud Resources

Artificial intelligence (AI) is emerging as a transformative force in cloud product optimization, enabling organizations to achieve efficiency, scalability, and cost-effectiveness that were previously unattainable. The complexity of managing cloud resources, including cost, performance, and reliability, has increased dramatically with the widespread adoption of cloud computing. AI techniques, such as machine learning, deep learning, and reinforcement learning, can be leveraged to address these complexities by predicting workload patterns, automating resource allocation, and ensuring optimal performance through proactive monitoring and adjustments. This article provides an in-depth exploration of AI-based methods used for optimizing cloud infrastructure, focusing on real-world scenarios like dynamic resource allocation, pricing prediction, and service reliability. Additionally, we present the challenges of AI adoption in cloud optimization and outline potential directions for future research

Research on short-term precipitation forecasting method based on CEEMDAN-GRU algorithm

Precipitation forecasting is vital for managing disasters, urban traffic, and agriculture. This study develops an improved model for short-term precipitation forecasting by combining Complete Ensemble Empirical Mode Decomposition with Adaptive Noise (CEEMDAN) and Gated Recurrent Unit (GRU). Using precipitation data from January 1, 2019, to December 31, 2022, as a sample, the model capitalizes on CEEMDAN’s superior signal decomposition capabilities and GRU’s ability to capture nonlinear dynamic patterns in time series. To assess the model’s effectiveness, comparisons were conducted with 12 benchmark models, including CEEMDAN-LSTM, EMD-GRU, EMD-LSTM, BI-LSTM, GRU, LSTM, and TCN. The results demonstrate that the CEEMDAN-GRU model achieves higher accuracy and stability in short-term precipitation forecasting. Leveraging an Adam optimizer with adaptive learning rate reduction enhances convergence and ensures reliable predictions, achieving an R²of 0.7915, MAE of 0.05382, and MSE of 0.09081.