Test Data Research Articles

Conversion Surgery After Chemotherapy Plus Nivolumab as the First-Line Treatment for Unresectable Advanced or Recurrent Gastric Cancer and a Biomarker Study Using the Gustave Roussy Immune Score: A Multicenter Study.

There are few reports on conversion surgery (CS) after chemotherapy plus nivolumab as a first-line treatment in patients with unresectable advanced or recurrent gastric cancer (GC). This multicenter study was conducted to analyze real-world data on CS after chemotherapy plus nivolumab as a first-line treatment and to identify predictive biomarkers. This multicenter study included 104 patients who received chemotherapy plus nivolumab as primary treatment for unresectable advanced recurrent GC from 12 institutes. We investigated and analyzed patient characteristics and blood test data in the presence or absence of CS, the relationship between the Gustave Roussy Immune Score (GRIm-s) and CS, and the characteristics of CS cases. CS was performed in 12 patients (11.5%). Eastern Cooperative Oncology Group Performance Status (ECOG-PS) was significantly better in patients who underwent CS (p<0.0001). There were no CS cases with high-risk GRIm-s (0%), however there were 22 non-CS cases (23.9%). No high-risk GRIm-s cases were converted to CS. Minimally invasive surgery was performed in 50.0% of the cases, with R0 resection in all cases and only one case of urinary retention (Grade II) as a postoperative complication, indicating a good postoperative short-term outcome. There were two cases of postoperative recurrence (16.7%), both of which were grade 1b. The short-term postoperative results of CS after chemotherapy plus nivolumab as the first-line treatment for GC were acceptable in this study. There were no high-risk GRIm-s cases among those who underwent CS, suggesting that the GRIm-s may be a predictor of CS.

Designing a reliable machine learning system for accurately estimating the ultimate condition of FRP-confined concrete

Precisely forecasting how concrete reinforced with fiber-reinforced polymers (FRP) responds under compression is essential for fine-tuning structural designs, ensuring constructions fulfill safety criteria, avoiding overdesigning, and consequently minimizing material expenses and environmental impact. Therefore, this study explores the viability of gradient boosting regression tree (GBRT), random forest (RF), artificial neural network-multilayer perceptron (ANNMLP) and artificial neural network-radial basis function (ANNRBF) in predicting the compressive behavior of fiber-reinforced polymer (FRP)-confined concrete at ultimate. The accuracy of the proposed machine learning approaches was evaluated by comparing them with several empirical models concerning three different measures, including root mean square errors (RMSE), mean absolute errors (MAE), and determination coefficient (R2). In this study, the evaluations were conducted using a substantial collection of axial compression test data involving 765 circular specimens of FRP-confined concrete assembled from published sources. The results indicate that the proposed GBRT algorithm considerably enhances the performance of machine learning models and empirical approaches for predicting strength ratio of confinement (f′cc/f′co) by an average improvement in RMSE as 17.3%, 0.65%, 66.81%, 46.12%, 46.31%, 46.87% and 69.94% compared to RF, ANNMLP, ANNRBF, and four applied empirical models, respectively. It is also found that the proposed ANNMLP algorithm exhibits notable superiority compared to other models in terms of reducing RMSE values as 9.67%, 11.29%, 75.11%, 68.83%, 73.64%, 69.49% and 83.74% compared to GBRT, RF, ANNRBF and four applied empirical models for predicting strain ratio of confinement (εcc/εco), respectively. The superior performance of the GBRT and ANNMLP compared to other methods in predicting the strength and strain ratio confinements is important in evaluating structural integrity, guaranteeing secure functionality, and streamlining engineering plans for effective utilization of FRP confinement in building projects.

EVALUASI ANALISIS GEOTEKNIK PADA TAMBANG TERBUKA BATUBARA PIT NORTH WEST PT ANAK TAMBANG INDONESIA JOB SITE PT ADIMITRA BARATAMA NUSANTARA, KALIMANTAN TIMUR

On a mine slope there will be a driving force and a resisting force. The mine slope will collapse if the driving force is greater than the resisting force. Thus, the principle of mitigating mine slope failure is to reduce the driving force or increase the resisting force. PT Adimitra Baratama Nusantara intends to develop a new mining area in the North-West area within the IUP area. Therefore, it is necessary to carry out a geotechnical study that focuses on slope stability analysis to support open-pit coal mining plans based on data from geotechnical investigations in the field, secondary data and geotechnical laboratory test data. Geotechnical modeling for slope stability analysis was carried out using the Finite Element Method because the overburden and interburden sedimentary rocks of coal seams are considered to have behavior approaching elasto-plastic, so that it will provide results that are close to the actual situation in the field. Modeling and slope stability analysis of the mine opening design represented by section-01 and section-03 with the height of each opening being 126 m and 119 m shows that the high-wall and low-wall slope designs are not yet stable, so a re-design is necessary. with the mine opening floor simulated at level -110 and level -90.

Evaluation of memory drought stress effects on storage compounds seedlings of cotton (Gossypium hirsutum) and in-silico analysis of glutathione reductase

In breeding programs, stress memory in plants can develop drought stress tolerance. Memory stress, as an approach, can keep stress data by activating tolerance mechanisms. This research was conducted to evaluate some physiologically effective mechanisms in inducing memory drought stress in the seeds that were exposed to water stress three times in four treatments including rainfed, 33%, 66%, and 100% of field capacity (FC). After the production of the seeds, the third-generation seeds were placed under different irrigation treatments, seed and seedling traits, starch to carbohydrate ratio in seed, protein concentration and glutathione reductase were investigatied in a factorial format based on a randomized complete block design with three replications. Results showed that percentage of changes from the lowest to the highest value for traits including seed vigor, seed endosperm weight, seed coat weight, accelerated aging, cold test, seedling biomass and seedling length were 25, 37, 65, 65, 55, 77, 55, 65 and 79, respectively and germination uniformity was 3.9 times higher than the lowest amount. According to the deterioration percentage, seed vigor and the percentage of seed germination in cold test data, it can be reported that seed production by 100% FC was not appropriate for rainfed plots. However, considering the the appropriate results in the percentage of germination for a cold test, germination uniformity percentage, and the lowest accelerated aging seeds, seed production under the rainfed conditions with 33% FC watering can be recommended. In-silico analysis was coducted on Glutathione reductase (GR) enzymes in Gossypium hirsutum. It is clear that GR has a Redox-active site and NADPH binding, and it interacts with Glutathione S transferase (GST). So, memory drought stress through inducing physiological drought tolerance mechanisms such as starch-to-carbohydrate ratio and GR can determine the suitable pattern for seed production for rainfed and low rainfall regions in a breeding program. Our study thus illustrated that seed reprduction under 33% FC equipped cotton with the tolerance against under draught stress from the seedling stage. This process is done through activating glutathione reductase and balancing the ratio of starch to carbohydrates concentration.

Numerical analysis of duct wall failure in a prototype FAIDUS design-based generation IV SFR fuel subassembly under flow blockage accident

An ingenious prototype Fuel Assembly Inner DUct Structure (FAIDUS) design is conceptualized under the Generation IV safety philosophy to prevent large-scale corium pool formation following total coolant flow blockage accident. The early failure of the duct wall provides a built-in guideway for molten material relocation out of the core region, ensuring inherent safety robustness. Therefore, understanding the duct wall failure mechanism is critical for evaluating accident mitigation potential of the FAIDUS design. In this regard, a transient enthalpy-based multiphase computational model is developed to investigate the heat transfer and fluid flow behavior involved during the duct wall failure in 3-D domain. The numerical model is employed initially to simulate wall failure in the EAGLE ID1 in-pile test. The model results are validated against the in-pile test data. Subsequently, FAIDUS duct wall failure in a prototype fast reactor fuel subassembly (SA) is investigated employing the present model. The study reveals that the large heat flux from the molten pool to the duct is the main reason for the duct wall failure. The duct failure starts with an initial rupture across the corners of the hexagonal-shaped duct and progresses to total failure of the duct wall by the expansion of the rupture over the whole duct wall area with time. The outer hexcan wall sustains only minor thermal damage during the duct wall failure process. The present study explores the associated thermal-hydraulics, molten pool dynamics, crust formation behavior, and event sequences involved in duct wall failure in great detail. The simulation findings suggest that the FAIDUS SA design facilitates premature failure of the slim duct wall prior to the hexcan wall failure. This could potentially enable early discharge of the molten corium away from the core region, limiting the accident progress.

基于蚯蚓运动特征的仿生筛清选性能试验研究

As the feeding mass of harvester continues to be increased, the accumulation of the maize mixture on the screen more serious, leading to a decrease in screening efficiency of the cleaning screen. The maize mixture on the screen was quickly transported backward through a bionic screen based on the motion characteristics of earthworms to reduce accumulation. However, the impact of the main operating parameters of this bionic screen on the cleaning performance was not yet clear. Therefore, the main operating parameters that affect the loss percentage and impurity percentage of maize grain in the cleaning device of bionic screen were simulated using CFD-DEM. Analysis of variance, regression and response surface were carried out on the test data, and the optimal parameter combination was obtained through parameter optimization: the concave depth of the screen surface was 25 mm, the installation angle of the screen surface was –5.5°, and the airflow velocity at the inlet of the cleaning device was 12.3 m/s. At this time, the lost percentage of maize grain was 1.56%, and the impurity percentage of maize grain was 1.85%, which met the requirements of national standards. The experiment provides reference and basis for the follow-up exploration of the screening mechanism of bionic screen.

An attempt to generate panoramic radiographs including jaw cysts using StyleGAN3.

The purpose of this study was to generate radiographs including dentigerous cysts by applying the latest generative adversarial network (GAN; StyleGAN3) to panoramic radiography. A total of 459 cystic lesions were selected, and 409 images were randomly assigned as training data and 50 images as test data. StyleGAN3 training was performed for 500 000 images. Fifty generated images were objectively evaluated by comparing them with 50 real images according to four metrics: Fréchet inception distance (FID), kernel inception distance (KID), precision and recall, and inception score (IS). A subjective evaluation of the generated images was performed by three specialists who compared them with the real images in a visual Turing test. The results of the metrics were as follows: FID, 199.28; KID, 0.14; precision, 0.0047; recall, 0.00; and IS, 2.48. The overall results of the visual Turing test were 82.3%. No significant difference was found in the human scoring of root resorption. The images generated by StyleGAN3 were of such high quality that specialists could not distinguish them from the real images.

Analysis of the frequency–response stability and reliability of a tower-column unstable rock mass on a high and steep slope

The first-order second-moment checking point method was introduced to judge the instability probability and evaluate the stability of the TCURM. The frequency–response stability calculation and the reliability results were compared, and a frequency–response stability and reliability analysis method was proposed. Taking the Zengziyan W12# unstable rock mass in Nanchuan, Chongqing, China, as an example, the calculation shows that the dynamic indexes and geological indexes decrease as the stiffness of the deterioration area decreases. According to the statistical data of the laboratory test and the field investigation, reliability theory is used to evaluate the stability of the TCURM, and the failure probabilities are 80.3% and 96.27% under natural and saturated conditions, which correspond to states of poor stability and instability, respectively. The reliability evaluation results are consistent with the conclusion of the frequency–response stability analysis. The new method can provide a theoretical basis for developing the dynamic monitoring and early warning indicators of the TCURM and disaster prevention and mitigation in mountainous areas.

Executive function tests: Normative data for Spanish-speaking adults in the United States

BACKGROUND: Normative data for Spanish-speaking populations, particularly Hispanics in the U.S., is notably scarce. OBJECTIVE: This study aims to establish normative data for executive function tests (Modified Wisconsin Card Sorting Test and Stroop Color and Word Test) among Spanish-speaking Hispanics in the U.S. METHODS: We assessed 245 individuals aged 18–80 from eight U.S. states (California, Connecticut, Florida, Indiana, New Jersey, Oregon, Virginia, and Wisconsin) and employed Bayesian regression to estimate norms, considering various sociodemographic factors influencing performance. RESULTS: The posterior distribution suggests a high probability that age impacts SCWT performance, with older adults likely to show greater declines, particularly among those with high proficiency in Spanish. The posterior distribution suggests a stronger effect of age on M-WCST performance among individuals with longer U.S. residency. Educational attainment demonstrates a robust positive impact on M-WCST outcomes, with lower levels of education associated with a higher probability of increased errors. An interaction between education and Spanish proficiency was observed, influencing SCWT scores differently across proficiency levels. Sex and acculturation levels interact to affect SCWT performance, with distinct patterns observed between men and women. This suggests that the impact of acculturation on cognitive test performance may vary by gender. CONCLUSION: Establishing culturally sensitive normative data can enhance accurate identification of executive dysfunction and reduce misdiagnosis risks. This study underscores the importance of considering sociocultural factors including acculturation and language proficiency in neuropsychological assessments to better serve diverse populations.

Determination of Aquifer Hydraulic Properties, Aquifer Potential and Vulnerability Characteristics Using Pumping Test and Surface Geophysical Data: A Case Study of Udi and Ezeagu Areas, Southeastern Nigeria

Determination of Aquifer Hydraulic Properties, Aquifer Potential and Vulnerability Characteristics Using Pumping Test and Surface Geophysical Data: A Case Study of Udi and Ezeagu Areas, Southeastern Nigeria

Effect of vibration on the interface properties of welded steel joints and filled concrete in steel pipes

Concrete-filled steel tubes (CFSTs) have been increasingly utilized in engineering due to their excellent mechanical properties. Ensuring a solid bond between a steel tube and concrete is essential for optimizing their synergistic effect. This study introduces an internally welded steel bar structure within the inner wall of a steel tube to enhance the bond properties at the connection interface. The influence of various configurations of steel bars welded to the inner surface of the tube on the bond strength is investigated considering the impact of vibration on the load-bearing capacity of the component. This study comprises two groups of specimens, one with vibration and one without vibration, for a total of ten specimens. Each group included CFST members with five distinct internal welded steel bar structures. The experimental results, including load–displacement curves and strain data of the steel tube, were used to assess the impact of the internal welded steel bar configurations on the steel–concrete interface. The sliding process is described by correlating test data with curves and observed phenomena. To comprehensively compare the effects of structural dimensions on the bonding and slipping properties of the welded bars, finite element simulations replicating the experimental conditions were carried out using ABAQUS software, and the simulation results agreed with the experimental observations. The study demonstrated that incorporating internal welded steel bars substantially enhances the bond strength of steel pipe–concrete interfaces. While vibration weakens the bond strength in CFST members, internal welded steel bars mitigate this effect. These findings improve the structural performance of CFST structures and their resilience to external vibrations.

Nondestructive Fatigue Life Prediction for Additively Manufactured Metal Parts through a Multimodal Transfer Learning Framework

Understanding the fatigue behavior and accurately predicting the fatigue life of laser powder bed fusion (L-PBF) parts remain a pressing challenge due to complex failure mechanisms, time-consuming tests, and limited fatigue data. This study proposes a physics-informed data-driven framework, a multimodal transfer learning (MMTL) framework, to understand process-defect-fatigue relationships in L-PBF by integrating various modalities of fatigue performance, including process parameters, XCT-inspected defects, and fatigue test conditions. It aims to leverage a pre-trained model with abundant process and defect data in the source task to predict fatigue life nondestructively with limited fatigue test data in the target task. MMTL employs a hierarchical graph convolutional network (HGCN) to classify defects in the source task by representing process parameters and defect features in graphs, thereby enhancing its interpretability. The feature embedding learned from HGCN is then transferred to fatigue life modeling in neural network layers, enabling fatigue life prediction for L-PBF parts with limited data. MMTL validation through a numerical simulation and real-case study demonstrates its effectiveness, achieving an F1-score of 0.9593 in defect classification and a mean absolute percentage log error of 0.0425 in fatigue life prediction. MMTL can be extended to other applications with multiple modalities and limited data.

Formulasi Sabun Cair Berbahan Baku Virgin Coconut Oil VCO dengan Penambahan Ekstrak Daun Pepaya Carica Papaya L. untuk Peningkatan Aktivitas Anti Bakteri Staphylococcus Epidermidis ATCC 12238

Acne is a chronic inflammatory disease caused by Staphylococcus epidermidis. Papaya leaves (Carica papaya L.) contain antibacterial substances, namely alkaloids carpaine, flavonoids, tocophenols. Of this study was to determine the physical quality of liquid soap, increasing antibacterial activity. Papaya leaves were extracted with 96% ethanol solvent, then evaporated with a water bath. Antibacterial activity test using the disc diffusion method in concentrations of 5%, 15% and 25% and liquid soap will be made F0 0%, F1 5%, F2 10%, and F3 25% against Staphylococcus epidermidis ATCC 12238. Amoxicillin tab as a positive control and distilled water as a negative control. Bacterial activity appears a clear zone around the disc paper. And see the physical quality test according to the requirements. The results of the bacterial test data are made in table form, and analyzed using the SPSS method Formula with the addition of papaya leaf extract can be made and according to the physical quality conditions of liquid soap. The results of the 25% concentration extract test are an inhibition zone of 20.8 mm, the highest liquid soap formula is an inhibition zone of 23.8 mm in inhibiting the growth of Staphylococcus epidermidis ATCC 12238. Liquid soap preparations and papaya leaf extract have antibacterial activity against Staphylococcus epidermidis ATCC 12238 bacteria and according to the physical quality requirements of liquid soap.

Development and validation of a questionnaire to assess the implementation of physical education programs in Chinese junior high schools

BackgroundStudents’ physical fitness has always been the focus of attention of the Chinese government, and the school as an important way to improve students’ physical fitness, there are many studies on the current status of the implementation of physical education in schools, and there are many studies that use self-made questionnaires to investigate the implementation of physical education in schools, but most of the studies do not adequately validate the self-made questionnaires, so the purpose of this study was to develop a questionnaire to assess the level of implementation of physical education programmes in Chinese junior secondary schools and to test its reliability and validity.MethodThe content of the questionnaire was developed based on the content of Annex 1 of the Assessment Measures for Physical Education in Primary and Secondary Schools issued by the Ministry of Education of China in 2014 and was modified based on feedback from the expert panel and pre-test participants. The questionnaire was initially tested for validity by 5 expert reviewers, and then we collected data information from 350 participants and conducted exploratory factor analysis (EFA) to explore the factor structure of the initial version. One week later, 40 of the 350 participants were randomly selected to assess test-retest reliability.ResultsThe I-CVI and KAPPA value analysis results of the expert review results show that the questionnaire has extremely high reliability and consistency among experts. EFA results indicate that the five dimensions of this questionnaire are highly reliable. In the test-retest reliability, the Pearson correlation coefficients of the initial test data and the retest data of each dimension are all greater than 0.7, and the significance probability values are all less than 0.05, reaching the significance level, the results show that the questionnaire has good stability.ConclusionsThis study concluded that the 5 dimensions and 38 items of this questionnaire had high reliability and validity and could be used as a preliminary tool to measure the implementation level of physical education programs in junior high schools in China. However, future research should explore the potential need for adjustment to suit different regions and cultures.

Optimization of the convolutional neural network classification model under the background of innovative art teaching models

To improve students’ ability to recognize and appreciate artworks, and further enhance their academic performance and classroom satisfaction, this study explores the application of the Convolutional Neural Network (CNN) model based on optimization in art teaching. Firstly, the importance and challenges of art teaching are analyzed. Secondly, the principle and structure of CNN and its application in the classification field are expounded, and then the CNN classification model is optimized. Finally, the effectiveness of the optimized model is verified by experiments. Experimental results show that the optimized model’s accuracy is up to 95.2% in the performance evaluation. The training time of the optimized model is much lower than that of the traditional model, and this model still maintains 95.2% accuracy under the noise of 14.7%. In addition, the accuracy of the optimized model on the unseen test data is 92%. In comparing teaching experiment results, by introducing the CNN classification model, Class B students’ average score of art homework has increased by 4.3 points. The score for class satisfaction is 8.1 points. This indicates that the optimized CNN model has significant advantages in art teaching and can effectively improve students’ classroom satisfaction and academic performance. Therefore, this study has specific reference significance for the innovation of the art teaching model.

Research of the calculation formula for working clearance leakage flow of the hydrogen circulation pump

The hydrogen circulation pump (HCP), a device for recovering unconsumed hydrogen gas in fuel cell systems to improve efficiency, is an important equipment in fuel cell systems. Efficient calculation of the output flow rate of the HCP is crucial for accelerating the product development process, but there is a lack of an effective calculation formula for the working clearance leakage. In this paper, a series of HCP models with different working clearances are established and calculated using an overlapping grid simulation method, verifying that the traditional Roots blower leakage flow formula is feasible for HCP, although the maximum calculation error reaches 10.71%. Further study the pressure distribution law inside the HCP chamber, and revise the traditional calculation formula accordingly, so that the average calculation error of the series models is reduced from 5.75% to 3.82%, and the maximum error is also reduced to 7.73%. Compared with the prototype test data, though the flow values obtained from the two calculation formulas are slightly higher, the calculation error of the correction formula is relatively smaller. The research results indicate that the correction formula can more accurately predict the flow rate of HCP and has important application value.

Leveraging Bayesian methods for addressing multi-uncertainty in data-driven seismic liquefaction assessment

When assessing seismic liquefaction potential with data-driven models, addressing the uncertainties of establishing models, interpreting cone penetration tests (CPT) data and decision threshold is crucial for avoiding biased data selection, ameliorating overconfident models, and being flexible to varying practical objectives, especially when the training and testing data are not identically distributed. A workflow characterized by leveraging Bayesian methodology was proposed to address these issues. Employing a Multi-Layer Perceptron (MLP) as the foundational model, this approach was benchmarked against empirical methods and advanced algorithms for its efficacy in simplicity, accuracy, and resistance to overfitting. The analysis revealed that, while MLP models optimized via maximum a posteriori algorithm suffices for straightforward scenarios, Bayesian neural networks showed great potential for preventing overfitting. Additionally, integrating decision thresholds through various evaluative principles offers insights for challenging decisions. Two case studies demonstrate the framework's capacity for nuanced interpretation of in situ data, employing a model committee for a detailed evaluation of liquefaction potential via Monte Carlo simulations and basic statistics. Overall, the proposed step-by-step workflow for analyzing seismic liquefaction incorporates multifold testing and real-world data validation, showing improved robustness against overfitting and greater versatility in addressing practical challenges. This research contributes to the seismic liquefaction assessment field by providing a structured, adaptable methodology for accurate and reliable analysis.

A modified bond-based peridynamic approach for rigid projectile perforation on concrete slabs

Peridynamic (PD) has a unique advantage in describing the crack growth and fragmentation of brittle materials. Concerning the dynamic behaviors and failure patterns of concrete slabs under projectile perforations, a modified bond-based PD approach maintaining both the easy implementation and computational stability characteristics was firstly developed from the following three aspects, (i) a rate-dependent PD constitutive model was proposed for describing the dynamic behaviors of concrete; (ii) a progressive damage criterion considering the tension-compression anisotropy, softening behavior, and strain rate effect of concrete was incorporated to more accurately reproduce the damage and failure of concrete; (iii) an improved micro-modulus function related to bond length was introduced to reveal the internal length effect of bond force. Then, numerical simulations of projectile perforation on concrete slabs by utilizing the developed modified bond-based PD approach, as well as the corresponding sensitivity analyses of discretization parameters including horizon size and particle spacing were performed. Based on the recommended horizon size and particle spacing, the predicted residual velocity of projectile and failure patterns of concrete slabs exhibited an excellent agreement with the test data. Furthermore, by comparisons of the traditional bond-based PD and classical finite element methods, the superiority of developed approach in describing the perforation damage of concrete targets against projectile impact was demonstrated. Finally, the modified bond-based PD approach was employed to blind simulate the projectile normal and oblique perforating multi-layered spaced concrete target plates. It was found that the modified PD model reasonably predicted the terminal ballistic trajectory, deflection angle, and residual velocity of projectile, as well as the failure patterns of target plates. The present work provides a new way to predict the terminal ballistic effect of projectile and dynamic behaviors of concrete slabs.

Transfer learning scenarios on deep learning for ultrasoundbased image segmentation

&lt;p&gt;Deep learning coupled with transfer learning, which involves reusing a pretrained model's network structure and parameter values, offers a rapid and accurate solution for image segmentation. Differing approaches exist in updating transferred parameters during training. In some studies, parameters remain frozen or untrainable (referred to as TL-S1), while in others, they act as trainable initial values updated from the first iteration (TL-S2). We introduce a new state-of-the-art transfer learning scenario (TL-S3), where parameters initially remain unchanged and update only after a specified cutoff time. Our research focuses on comparing the performance of these scenarios, a dimension yet unexplored in the literature. We simulate on three architectures (Dense-UNet-121, Dense-UNet-169, and Dense-UNet-201) using an ultrasound-based dataset with the left ventricular wall as the region of interest. The results reveal that the TL-S3 consistently outperforms the previous state-of-the-art scenarios, i.e., TL-S1 and TL-S2, achieving correct classification ratios (CCR) above 0.99 during training with noticeable performance spikes post-cutoff. Notably, two out of three top-performing models in the validation data also originate from TL-S3. Finally, the best model is the Dense-UNet-121 with TL-S3 and a 20% cutoff. It achieves the highest CCR for training 0.9950, validation 0.9699, and testing data 0.9695, confirming its excellence.&lt;/p&gt;

Stream salinity prediction in data-scarce regions: Application of transfer learning and uncertainty quantification

Scarcity of stream salinity data poses a challenge to understanding salinity dynamics and its implications for water supply management in water-scarce salt-prone regions around the world. This paper introduces a framework for generating continuous daily stream salinity estimates using instance-based transfer learning (TL) and assessing the reliability of the synthetic salinity data through uncertainty quantification via prediction intervals (PIs). The framework was developed using two temporally distinct specific conductance (SC) datasets from the Upper Red River Basin (URRB) located in southwestern Oklahoma and Texas Panhandle, United States. The instance-based TL approach was implemented by calibrating Feedforward Neural Networks (FFNNs) on a source SC dataset of around 1200 instantaneous grab samples collected by United States Geological Survey (USGS) from 1959 to 1993. The trained FFNNs were subsequently tested on a target dataset (1998-present) of 220 instantaneous grab samples collected by the Oklahoma Water Resources Board (OWRB). The framework's generalizability was assessed in the data-rich Bird Creek watershed in Oklahoma by manipulating continuous SC data to simulate data-scarce conditions for training the models and using the complete Bird Creek dataset for model evaluation. The Lower Upper Bound Estimation (LUBE) method was used with FFNNs to estimate PIs for uncertainty quantification. Autoregressive SC prediction methods via FFNN were found to be reliable with Nash Sutcliffe Efficiency (NSE) values of 0.65 and 0.45 on in-sample and out-of-sample test data, respectively. The same modeling scenario resulted in an NSE of 0.54 for the Bird Creek data using a similar missing data ratio, whereas a higher ratio of observed data increased the accuracy (NSE = 0.84). The relatively narrow estimated PIs for the North Fork Red River in the URRB indicated satisfactory stream salinity predictions, showing an average width equivalent to 25 % of the observed range and a confidence level of 70 %.