Linear Machine Research Articles

A Novel Efficiency Optimization Method for Linear Oscillatory Machine Based on Variable Universe Fuzzy Search

A Novel Efficiency Optimization Method for Linear Oscillatory Machine Based on Variable Universe Fuzzy Search

Forecasting Crude Oil Prices with a Structural Machine Learning Model

This study proposes a structural machine learning methodology that integrates both linear and nonlinear relationships for crude oil price forecasting. By employing a partially linear machine learning model that explicitly captures the linear effects of key variables influencing crude oil prices, the approach enhances interpretability while evaluating predictive performance. In addition, this study investigates the impact of hyperparameter selection on forecasting accuracy, with a particular emphasis on subsampling and random seed effects–factors that have received limited attention in existing empirical research. Subsampling is actively utilized as a hyperparameter to explore variations in predictive performance, and instead of relying on a single fixed random seed, multiple seeds are used to assess the model's stability and robustness. Based on monthly forecasting experiments spanning approximately 11 years, the results demonstrate that the partially linear machine learning model, when optimized through appropriate subsampling and hyperparameter tuning, outperforms benchmark models in one- to three-step-ahead forecasts. Furthermore, an analysis of prediction error distributions across different random seeds confirms the robustness of the model's predictive performance.

Data-Driven Prediction of Binder Rheological Performance in RAP/RAS-Containing Asphalt Mixtures

Asphalt recycling technologies have advanced considerably over the last few decades with the utilization of reclaimed asphalt pavements (RAP) and recycled asphalt shingles (RAS). Characterizing aged and heterogeneous binders in these mixtures is challenging, particularly with limited extracted binders. This study suggests a data-driven framework that considers the rheological, chemical, and thermal characteristics to predict the binders’ performance. Ninety-seven mixtures with 0–35% of the asphalt binder replaced with RAP/RAS binders were included as cores from the field, plant-produced mixtures, and laboratory-fabricated mixtures. The binders were chemically quantified using aging, aromatic, and aliphatic indices. Thermal analyses of the binders involved the percentage of the thermal residue. The framework predicted the rheological resistance of the binders to rutting and cracking using linear and nonlinear machine learning models. The nonlinear models outperformed the linear models for the three rheological parameters. The nonlinear models achieved a 69% reduction in the root mean square error (RMSE) for rutting, a 37% reduction in the RMSE for fatigue cracking, and a 21% reduction in the RMSE for thermal cracking. However, the nonlinear models overfitted for block cracking and had an RMSE 41% higher than the linear models, despite a perfect correlation (R = 1.00). The feature importance demonstrated the strong effects of the chemical and thermal parameters on rheological prediction. The data-driven framework can successfully support efforts to better manage asphalt recycling by predicting the binder performance.

Harnessing predictive analytics to support high-risk learners in a one-year certification program in emergency medicine (CPEM) in Pakistan

Introduction Predictive analytics and Machine Learning (PAML) are gaining traction in health professions education (HPE). Their utilization includes, but is not limited to, guiding student enrollment, identifying at-risk learners, enhancing educational decisions, and allocating proper resources through data-driven insights. This study explored the use of PAML to identify at-risk learners in a one-year Certification Program in Emergency Medicine (CPEM) at the Indus Hospital and Health Network (IHHN), Pakistan with the aim of providing targeted educational support for improved outcome. Methodology By leveraging data from prior CPEM cohorts (2018–2022, n = 91), regression tree and linear regression machine learning models were compared to predict the final examination performance of the CPEM 2023 learner cohort (n = 26). The models were prospectively applied to identify at-risk learners (n = 14/26). Extra learning support (ELS) was offered as an inclusive measure to everyone, not just the ones flagged by the models and was accepted by ten learners. Data were analyzed for model accuracy and the impact of the educational intervention. Results Both models showed high accuracy (regression tree: Area Under the Receiver Operating Characteristic (ROC) Curve (AUC)= 0.89; linear regression: AUC= 0.88), though the regression tree model demonstrated slightly better sensitivity and specificity. The models altogether predicted unsatisfactory performance for 14 learners scheduled to sit for the 2023 final examination. Following targeted intervention, eight learners showed improvement in their final scores. Regression tree model was comparatively better in making predictions; however, both models had their limitation. Conclusion The study demonstrated the feasibility and utility of using PAML to identify at-risk learners and tailor support strategies for enhancing educational outcome in low-resource settings. This additional support can augment expert judgement and ensure equitable educational practices. However, model limitations and ethical concerns, such as algorithmic bias, overfitting, and data imbalance, must be actively addressed in high-stakes assessments. Practice points Predictive analytics and machine learning (PAML) can be used to generate early, data-informed signals that support proactive intervention. Machine learning algorithms capture complex, non-linear relationships between learner characteristics. Real time data integration and faculty feedback can enhance model accuracy. Ethical, inclusive, and targeted interventions provide tailored learning support. Bias mitigation and regular model validation ensure maximal utilization of limited resources.

Digital mapping of soil erodibility factor in response to land use change using machine learning models

Understanding the spatial variability of soil erodibility and its associated indices across different land uses is critical for sustainable land use planning and management. Traditional methods for measuring these variables are often time-consuming and costly. To address this, the study employed digital soil mapping (DSM) and machine learning (ML) models as efficient and cost-effective alternatives to predict soil erodibility and its indices, including clay ratio, critical level of organic matter, crust formation, dispersion ratio, and soil aggregate stability. 50 soil surface samples (0–20 cm depth) were collected from forest, agricultural, and pasture land uses. Soil physicochemical properties were determined through laboratory analyses. The study utilized Multiple Linear Regression (MLR) and machine learning models, including Random Forest (RF), Support Vector Machine (SVM), Artificial Neural Network (ANN), and an ensemble of the four single models. These models were trained using the repeated tenfold cross-validation method and evaluated based on root mean square error (RMSE), mean absolute error (MAE), and coefficient of determination (R2). The results demonstrated that the ANN model outperformed others in predicting soil erodibility (R2 = 0.98, MAE = 0.00341, RMSE = 0.0031. The SVM and RF models also performed well, with SVM achieving R2 = 0.93, MAE = 0.00541, RMSE = 0.0038, and RF achieving R2 = 0.87, MAE = 0.0037, RMSE = 0.00557 for soil erodibility prediction. The superior performance of ANN is attributed to its ability to model complex, non-linear interactions among variables influencing soil erodibility. Nonetheless, challenges such as data quality requirements and the risk of overfitting highlight the need for careful model calibration. The spatial prediction of soil erodibility across land uses revealed distinct patterns. Forest soils exhibited the lowest mean erodibility values (0.0313 t ha⁻1 h MJ⁻1 mm⁻1), reflecting their higher resistance to erosion due to better soil structure and organic matter content. In contrast, agricultural land uses recorded the highest mean erodibility values (0.0320 t ha⁻1 h MJ⁻1 mm⁻1), likely due to frequent tillage and reduced vegetation cover, which increase erosion susceptibility. Among soil types, Calcaric Cambisols were identified as the most erosion-prone, while Lithic Leptosols were the least susceptible, attributed to differences in soil texture, structure, and organic matter content. Finally, the basin was classified based on soil erodibility classes. The analysis showed that 81.18% of the basin (covering 546.6 km2) falls under the less erodible class, highlighting the basin’s overall resilience to erosion. In conclusion, the study demonstrates that machine learning-based models can accurately predict soil erodibility and its indices. The resulting maps provide a valuable baseline for land use planning, natural resource management, and decision-making processes.

QSAR Model Development for the Environmental Risk Limits and High-Risk List Identification of Phenylurea Herbicides in Aquatic Environments.

Due to the extensive residues of phenylurea herbicides (PUHs) in the environment, it is important for the ecological risk assessment of PUHs to determine their environmental risk limits and identify the high-risk PUHs. This study derived the environmental risk limit (HC5) of PUHs based on the species sensitivity distribution method and obtained the molecular descriptors using the ORCA and Dragon software. Based on the derived HC5 and the molecular descriptors, quantitative structure-activity relationship (QSAR) models were developed to predict the HC5 values using multiple linear regression (MLR) and machine learning (ML) methods. Then, the ecological risk assessment was carried out based on the monitored environmental concentration and the predicted HC5, and a list of high-risk PUHs was proposed. The results indicated that the derived HC5 concentrations of 36 PUHs vary greatly, ranging from 0.0000084963 to 5.1512 mg/L. The performance of both the developed QSAR models by the MLR and RF methods met the OECD requirements. Comparatively, the RF model showed a better predictive performance, with a higher correlation coefficient between the experimental HC5 and predicted HC5 (R2 = 0.90) than the MLR model (R2 = 0.86). The developed QASR models also provided insights into the influence of the molecular descriptors on toxicity that the spatial structural descriptors, electronic descriptors, and hydrophobicity descriptors are key descriptors affecting the toxicity of PUHs. The high-risk PUH list from the ecological risk assessment demonstrated that the risk quotient of 10 PUHs (diuron, rimsulfuron, thifensulfuron-methyl, metsulfuron-methyl, metsulfuron, isoproturon, pyrazosulfuron, bensulfuron, tribenuron-methyl, and tebuthiuron) ranged from 4.39 to 2977.68, which are high-risk PUHs that should be given more attention. The obtained results can provide important basis for the ecological risk assessment of PUHs.

Comparing the Performance of Regression and Machine Learning Models in Predicting the Usable Area of Houses with Multi-Pitched Roofs

The usable floor area is one of the key parameters when appraising residential property. In Poland, valuers have to base their analysis on data from the Real Estate Price Register (RCN) in order to value a property. Unfortunately, these data often turn out to be incomplete, especially with regard to floor area, which makes the selection of reference properties difficult and can lead to erroneous valuation results. To address this problem, a study was conducted that used linear models, non-linear models and machine learning algorithms to calculate the floor area of buildings with complex multi-pitched roofs. The analysis was conducted using data sourced from the Database of Topographic Objects (BDOT10k). Three key factors were identified to provide a reliable estimate of usable floor area: the covered area, the height of the building and, optionally, the number of storeys. The results show that the linear model based on the design data achieved an accuracy of 88%, the non-linear model achieved 89% and the machine learning algorithms achieved 93%. For the existing building data from the city of Koszalin, the best model achieved an accuracy of 90%. The estimated values of the usable area of the building designs for the best model on the test set differed on average from the true ones by 8.7 m2, while for the existing buildings, the difference was 9.9 m2 on average (in both cases, the average relative error was about 7%).

Design and Multi-Objective Optimization of a Tubular Linear PM Machine for Stirling Generator Based on Gradient Boosting Regression Tree Surrogate Model

Design and Multi-Objective Optimization of a Tubular Linear PM Machine for Stirling Generator Based on Gradient Boosting Regression Tree Surrogate Model

Clinical nomogram for determining expected choroidal thickness in children with myopia.

Clinical nomogram for determining expected choroidal thickness in children with myopia.

Development of a High-Thrust Dual-Mover Transverse-Flux Linear Oscillatory Machine With Spoke PMs Embedded on Mover Yoke

Development of a High-Thrust Dual-Mover Transverse-Flux Linear Oscillatory Machine With Spoke PMs Embedded on Mover Yoke

Assessment of the availability of radiotherapy services across Nigeria.

e13572 Background: Radiotherapy is an important treatment mode in managing cancers, more so in Africa and Nigeria in particular, where most patients with cancers tend to present with advanced disease. Chemoradiation is the recommended treatment in advanced cervical cancer. The aim of this survey was to review the availability of radiotherapy facilities across Nigeria given the importance of cervical cancer as the second commonest cause of cancer death among Nigerian women. Methods: The list of all hospitals rendering cervical cancer care and the availability of radiotherapy machines; both Medical Linear Accelerator (LINAC) and High Dose Rate (HDR) brachytherapy, were reviewed across the six geopolitical zones in Nigeria. Results: Nigerian population is estimated be over 200 million with 44% of these been women. There are 33 teaching hospitals, 24 Federal Medical centres, 22 Specialist hospitals and hundreds of private hospitals. However, there are only seven Linear Accelerators and six Brachytherapy machines available for care of cancer patients in the country. Nigeria is divided into six Geopolitical zones: North-Central (NC), North-East (NE), North-West (NW), South-South (SS), South- West (SW) and South-East (SE). NC zone with over 46 million population, has only two LINACs and one Brachytherapy machine, located at the National Hospital, Abuja. NW zone with over 49 million population has two LINAC and one HDR brachytherapy at UDUTH Sokoto, Muhammadu Buhari Specialist Hospital Kano, and ABUTH Zaria respectively. NE with over 26 million population has two brachytherapy machines at Federal Teaching Hospital Gombe and the University of Maiduguri Teaching Hospital (UMTH) and a non-functional linear accelerator machine IN UMTH. SS zone with over 28 million population, has a single government-owned radiotherapy centre and one privately owned facility with a functional LINAC. SW zone with over 46 million population has one HDR in the University College Hospital (UCH), Ibadan. A Public-private partnership in Lagos University Hospital/NSIA has both LINAC and HDR while a privately owned centre (Marcellus Ruth Cancer Centre) has a LINAC. The SE zone with over 21 million population has one LINAC in the University of Nigeria (UNN), Nsukka. Majority of the government owned Radiotherapy services provide epileptic services as most of the machines are faulty leading to increased patient waiting time. Conclusions: Despite the high cancer burden in Nigeria, there are few radiotherapy machines leading to suboptimal and epileptic treatment of cervical cancer. The privately owned centres provide more reliable and prompter but expensive care. There is a need for more investment in cancer treatment and continued global support towards provision of radiotherapy machines and cancer treatment in general to reduce its associated morbidity and mortality in Nigeria and other LMICs.

Design optimization and generating characteristics of a linear arc PM vernier machine for wave energy conversion system

Design optimization and generating characteristics of a linear arc PM vernier machine for wave energy conversion system

The kinetics of epitope-specific IgE and IgG4 in early peanut allergy development and resolution.

The kinetics of epitope-specific IgE and IgG4 in early peanut allergy development and resolution.

Supply Chain Optimization in Manufacturing

The optimization of supply chain management (SCM) is crucial for enhancing efficiency and reducing costs in manufacturing industries. This study employs a qualitative research approach to explore various strategies for SCM optimization. Data were collected through semi-structured interviews with SCM experts, focus groups with key stakeholders, and an extensive review of secondary sources. The analysis reveals that the integration of information technology (IT), including ERP systems, supply chain management software, cloud computing, and the Internet of Things (IoT), significantly enhances supply chain visibility and coordination, leading to improved decision-making and reduced lead times. Additionally, lean manufacturing and Just-In-Time (JIT) practices are found to be effective in minimizing waste, optimizing inventory levels, and aligning production schedules with market demand, thereby reducing costs and increasing operational efficiency. Strategic supplier partnerships and collaborations play a vital role in achieving synchronization across the supply chain, improving quality, and managing risks. The adoption of sustainability and green supply chain management (GSCM) practices is also highlighted as a key driver for cost reduction and efficiency improvement. These practices not only enhance environmental performance but also drive innovation and provide a competitive advantage. This research paper explores advanced strategies and methodologies for optimizing supply chains within the manufacturing sector. It examines the integration of technologies such as Artificial Intelligence (AI), Internet of Things (IoT), and data analytics to streamline operations from raw material procurement to product delivery. The study highlights key optimization models, including linear programming, heuristic algorithms, and machine learning-based predictive models, while addressing real-world constraints like demand variability, production lead times, and logistics uncertainties. Furthermore, it discusses the role of sustainable practices in modern supply chain design, emphasizing the need for resilience and agility amidst global disruptions. Case studies from leading manufacturers illustrate successful implementation and the measurable impact of optimization efforts. This paper concludes with recommendations for future research directions focused on adaptive and autonomous supply chain systems.

Health-Related Quality of Life and Everyday Functioning in the Flood-Affected Population in Germany - A Case Study of the 2021 Floods in West Germany.

Floods lead to adverse impacts not only in financial terms but also on the health of the exposed population. We report on health-related Quality of Life (QoL) and functioning in the population affected by the 2021 flooding in Germany using an empirical survey data set. Health-related QoL and functioning are represented by two scores-(a) The EuroQoL 5D Visual Analog Scale (EQ-5D VAS) and (b) The 12-Item World Health Organization Disability Assessment Schedule (WHODAS 2.0), respectively. By applying an incremental linear regression model and Machine Learning models, we infer that health-related QoL and functioning are strongly negatively related to the psychological burden from those being affected by the flooding. This includes how often they think about the traumatic event. Home owners were found to have worse QoL and functioning than tenants. Household income and the status of repair/reconstruction of flood damages-in specific, insurance benefits, private donation and satisfactory claims compensation are associated with high health-related QoL and functioning. These findings highlight the importance of strengthening the health-related QoL of flood affected populations and emphasizes the strong association between recovery and health-related QoL and functioning of flood-affected populations.

Linear Discriminant Analysis-Based Machine Learning and All-Atom Molecular Dynamics Simulations for Probing Electro-Osmotic Transport in Cationic-Polyelectrolyte-Brush-Grafted Nanochannels.

Deciphering the correct mechanisms governing certain phenomena in polyelectrolyte (PE) brush grafted systems, revealed through atomistic simulations, is an extremely challenging problem. In a recent study, our all-atom molecular dynamics (MD) simulations revealed a nonlinearly large electroosmotic (EOS) flow (in the presence of an applied electric field) in nanochannels grafted with PMETAC [poly(2-(methacryloyloxy)ethyl trimethylammonium chloride] brushes. Given the lack of any formal procedure that would have directed us to identify the correct factors responsible for such an occurrence, we needed to devote several months to unraveling the involved mechanisms. In this letter, we propose a linear discriminant analysis (LDA)-based machine learning (ML) approach to address this gap. At first, we obtained data on certain basic features from the all-atom MD data. These basic features represent the number of atoms of a certain species around one atom of another (or the same) species. We obtain such data on basic features for a reference case (case of an EOS flow in PMETAC-brush-grafted nanochannels with a smaller electric field) and a perturbed case (case of an EOS flow in PMETAC-brush-grafted nanochannels with a larger electric field) in bins into which the nanochannel half height has been divided. These data sets are high-dimensional data sets to which the LDA is applied. This leads to the projection of the data (between the reference and the perturbed states) in a highly separated form on a 1D line. From such LDA calculations, we can identify the relative importance of the different basic features in ensuring this separation of the data (between the reference and the perturbed states) on the 1D line. The relative importance of the different basic features is quantified as "importance scores" for the different features, which, in turn, tell us what to study and where to study. Such knowledge enables us to rapidly identify the key factors responsible for the nonlinearly large EOS transport in PMETAC-brush-grafted nanochannels.

Deciphering the history of ERK activity from fixed-cell immunofluorescence measurements

The RAS/ERK pathway plays a central role in diagnosis and therapy for many cancers. ERK activity is highly dynamic within individual cells and drives cell proliferation, metabolism, and other processes through effector proteins including c-Myc, c-Fos, Fra-1, and Egr-1. These proteins are sensitive to the dynamics of ERK activity, but it is not clear to what extent the pattern of ERK activity in an individual cell determines effector protein expression, or how much information about ERK dynamics is embedded in the pattern of effector expression. Here, we evaluate these relationships using live-cell biosensor measurements of ERK activity, multiplexed with immunofluorescence staining for downstream target proteins of the pathway. Combining these datasets with linear regression, machine learning, and differential equation models, we develop an interpretive framework for immunofluorescence data, wherein Fra-1 and pRb levels imply long-term activation of ERK signaling, while Egr-1 and c-Myc indicate more recent activation. Analysis of multiple cancer cell lines reveals a distorted relationship between ERK activity and cell state in malignant cells. We show that this framework can infer various classes of ERK dynamics from effector protein stains within a heterogeneous population, providing a basis for annotating ERK dynamics within fixed cells.

Uncemented hip arthroplasty and denosumab: increased postoperative dipeptide concentrations and identification of potential new bone turnover biomarkers

Denosumab is a potent antagonist of RANKL and is widely used to treat severe postmenopausal osteoporosis. Using high-resolution mass spectrometry (HRMS), we aimed to identify molecular mediators associated with the rapid reactivation of osteoclasts following discontinuation of denosumab. In a previously reported randomized controlled trial, 64 patients undergoing uncemented total hip arthroplasty were randomized to receive 2 doses of 60 mg denosumab or placebo, administered 1-3 d and 6 mo postoperatively. Serum samples were analyzed using untargeted HRMS coupled with liquid chromatography, and bone turnover markers were assessed. Data were evaluated using linear mixed-effects models and machine learning techniques. After surgery, 83 metabolite features showed significant concentration changes (p < .0001). Denosumab-treated patients exhibited increased levels of the dipeptides di-L-phenylalanine, phenylalanylleucine, and alpha-Asp-Phe, and decreased levels of fibrinopeptide A and related peptides 24 mo after surgery. The oxidized peptide AP(Ox)GDRGEP(Ox)GPP(Ox)GP, derived from the collagen type I alpha 1 chain (COL1A1) and referred to as COL1A1-OxP, showed a strong correlation with the bone formation marker procollagen type 1 amino-terminal propeptide (P1NP) (p = 4.4E−83). Similarly, the tripeptide DL-alpha-aspartyl-DL-valyl-DL-proline (DVP) correlated highly with the bone resorption marker carboxy-terminal telopeptide of type 1 collagen (CTX) (p = 1.1E−222). P1NP and CTX levels were suppressed at 3, 6, and 12 mo postoperatively but exceeded baseline levels by 24 mo. Global metabolic shifts were observed postoperatively, with distinct profiles between treatment groups. The observed increase in specific dipeptides may reflect mechanisms contributing to rebound bone loss following denosumab withdrawal. Fibrinopeptide A and its analogs may play a protective role, while COL1A1-OxP and DVP represent potential new markers of bone turnover. These findings suggest metabolomics-based biomarkers could aid clinical decision-making by allowing earlier detection of rebound effects and guiding individualized treatment strategies after denosumab therapy.Clinical trial registration number: ClinicalTrials.gov, NCT01630941 (URL: https://clinicaltrials.gov/); European Union Clinical Trials Register (EU CTR), EudraCT No. 2011-001481-18 (https://www.clinicaltrialsregister.eu/)

Linear regression and machine learning modelling for chlorophyll content estimation using leaf red, green and blue images

Plant chlorophyll content is a key indicator of crop quality status. The conventional methods and procedures for determining plant chlorophyll content are laborious, timeconsuming and costly. However, with the adaptation to machine learning, plant data can be analysed more proficiently using red, green and blue (RGB) images obtained from a smartphone camera. Therefore, this study aimed to utilise machine learning algorithms to predict the chlorophyll content of lettuce based on RGB leaf images. Machine learning algorithms were run using RapidMiner software on indices of 60 images. The actual leaf chlorophyll content was measured using a SPAD chlorophyll meter. The correlation of the ratio between the green channel and red channel (GDR) indices with the leaf chlorophyll content, obtained using linear regression, was around 79.91%, with the lowest Root Mean Square Error (RMSE) of 6.62 g of chlorophyll/100 g fresh tissue. The use of machine learning algorithms with principal component analysis (PCA) increased the estimation accuracy by as much as 24%. The greatest accuracy was achieved using the Support Vector Machine (SVM) algorithm with selected highly correlated image indices, resulting in the lowest RMSE of 5.07 g of chlorophyll/100 g of fresh tissue.

Deep Artificial Neural Network Modeling of the Ablation Performance of Ceramic Matrix Composites in the Hydrogen Torch Test

In recent years, there has been increasing interest in new materials such as ceramic matrix composites (CMCs) for power generation and aerospace propulsion applications through hydrogen combustion. This study employed a deep artificial neural network (DANN) model to predict the ablation performance of CMCs in the hydrogen torch test (HTT). The study was conducted in three phases to increase the accuracy of the model’s predictions. Initially, to predict the thermal behavior of ceramic composites, two linear machine learning models were used known as Lasso and Ridge regression. In the second step, four decision tree-based ensemble machine learning models, namely random forest, gradient boosting regression, extreme gradient boosting regression, and extra tree regression, were used to improve the prediction accuracy metrics, including root mean square error (RMSE), mean absolute error (MAE), correlation coefficient (R2 score), and mean absolute percentage error (MAPE), relative to the previously introduced linear models. Finally, to forecast the thermal stability of CMCs with time, an optimized DANN model with two hidden layers having rectified linear unit activation function was developed. The data collection procedure involved preparing CMCs with continuous Yttria-Stabilized Zirconia (YSZ) fibers and silicon carbide (SiC) matrix using a polymer infiltration and pyrolysis (PIP) technique. The samples were exposed to a hydrogen flame at a high heat flux of 183 W/cm2 for a duration of 10 min. A good agreement between the DANN model’s predictions and experimental data with an R2 score of 0.9671, RMSE of 16.45, an MAE of 14.07, and an MAPE of 3.92% confirmed the acceptability of the developed neural network model in this study.