Lower Prediction Error Research Articles

Identifying Factors Affecting the Survival of Patients with HIV-Associated B-Cell Lymphoma Using a Random Survival Forest Model.

There have been no reports about the application of random survival forest (RSF) model to predict disease progression of HIV-associated B-cell lymphoma. A total of 44 patients with HIV-associated B-cell lymphoma who were referred to Nanjing Second Hospital from 2012 to 2019 were included. The RSF model was used to find predictors of survival, and the results of the RSF model were compared with those of the Cox model. The data were analyzed using R software (version 4.1.1). One-, 2-, and 3-year survival rates were 74.5%, 57.7%, and 48.6%, respectively, and the median survival was 59.0 months. The first 3 most important predictors of survival included lactate dehydrogenase (LDH), absolute monocyte count (AMC), and white blood cells (WBCs) count. The median survival of high-risk patients was only 4.0 months. Areas under the curve (AUCs) of the RSF model remained at more than 0.90 at 1, 2, and 3 years. The RSF model displayed a lower prediction error rate (21.9%) than the Cox model (25.4%). Lactate dehydrogenase, AMC, and WBCs count are the most important prognostic predictors for patients with HIV-associated B-cell lymphoma. Much larger prospective and/or multicentre studies are required to validtae this RSF model.

A portable dry film FTIR instrument for industrial food and bioprocess applications.

The main objective of this study was to design, build, and test a compact, multi-well, portable dry film FTIR system for industrial food and bioprocess applications. The system features dry film sampling on a circular rotating disc comprising 31 wells, a design that was chosen to simplify potential automation and robotic sample handling at a later stage. Calibration models for average molecular weight (AMW, 200 samples) and collagen content (68 samples) were developed from the measurements of industrially produced protein hydrolysate samples in a controlled laboratory environment. Similarly, calibration models for the prediction of lactate content in samples from cultivation media (59 samples) were also developed. The portable dry film FTIR system showed reliable model characteristics which were benchmarked with a benchtop FTIR system. Subsequently, the portable dry film FTIR system was deployed in a bioprocessing plant, and protein hydrolysate samples were measured at-line in an industrial environment. This industrial testing involved building a calibration model for predicting AMW using 60 protein hydrolysate samples measured at-line using the portable dry film FTIR system and subsequent model validation using a test set of 26 samples. The industrial calibration in terms of coefficient of determination (R2 = 0.94), root mean square of cross-validation (RMSECV = 194 g mol-1), and root mean square of prediction (RMSEP = 162 g mol-1) demonstrated low prediction errors as compared to benchtop FTIR measurements, with no statistical difference between the calibration models of the two FTIR systems. This is to the authors' knowledge the first study for developing and employing a portable dry film FTIR system in the enzymatic protein hydrolysis industry for successful at-line measurements of protein hydrolysate samples. The study therefore suggests that the portable dry film FTIR instrument has huge potential for in/at-line applications in the food and bioprocessing industries.

Comparison of Predictability in Vault Using NK Formula and KS Formula for the Implantable Collamer Lens Surgery.

This study aims to investigate the agreement between the NK and KS formulas in predicting the vault after implantation of an EVO-implantable collamer lens (ICL). This retrospective study included 106 eyes of 57 patients who underwent ICL-V4c implantation. Preoperative vault prediction was conducted by utilizing the NK and KS formulas, with postoperative measurements by anterior segment optical coherence tomography (AS-OCT) at one month. The analysis focused on the consistency between predicted and achieved vaults, as well as the correlation between the achieved vault and various biometric parameters. The mean achieved vault was 605.25 ± 212.72 µm, which was significantly smaller than the predicted vaults of 710.08 ± 195.08 and 673.80 ± 212.76 µm, using the NK and KS formulas, respectively (P < 0.05). The mean differences between the achieved vault and the predicted vault using the NK formula and KS formula were -104.82 μm (95% LoA: -600.38-391.19 μm) and -68.55 μm (95% LoA: -628.91-491.82 μm), respectively. Anterior chamber depth (ACD), vertical sulcus-to-sulcus (V-STS) diameter, and crystalline lens rise (CLR) were independent factors associated with the achieved vault (P < 0.05). The two formulas showed no statistically significant difference in absolute prediction error (APE). The NK formula exhibited superior consistency and low predictive error compared to the KS formula in the 12.6 mm ICL group. AS-OCT measurements overestimated the predicted ICL vault, especially in the 13.2 mm ICL size selection. Relying solely on the NK or KS formulas for predicting vaults before ICL surgery is not recommended.

Prediction model of a shell and tube heat exchanger based on the technique of artificial neural networks

The purpose of this paper is to report the proposal of a temperature prediction model methodology for a shell and tube heat exchanger using Artificial Neural Networks (RNAs). For the generation of the model, a set of historical data of five years was used, where 1633 readings from the equipment were obtained. The recorded data were the heat transfer coefficients and the fuel flow to predict the fluid temperature. The proposed methodology uses three stages. The first was the scaling of the data set between 0 and 1, this was done to facilitate the training of the RNA model: The second was to apply data mining techniques to create data clusters to model the behavior of the heat exchanger. The last stage was the evaluation of the prediction models. 5 proposals for Neural Network models were evaluated, these used 10 neurons in the hidden layer, the main difference between them was the number of clusters used in the training data that increased one by one. The average training errors obtained by the four types of data pools were 0.00220, 0.00190, 0.00133, 0.00098, and 0.00080. According to the results obtained, it was possible to conclude that the model that uses a greater number of clusters has a lower prediction error.

Modelling of the surface morphology and size effects on fatigue strength of L-PBF Inconel 718 by comparing different testing specimens

The aim of this study was to investigate the impact of surface roughness and size effect on the uniaxial fatigue strength of aged specimens made of Inconel 718, obtained by laser-powder bed fusion (L-PBF) both under as-built and machined surface conditions. The surface profiles were scanned with a 3D optical profilometer, and then modelled and filtered by the fast Fourier transform. This latter allowed to obtain the relevant parameters to calculate the term area in the Murakami model to account for the effect of surface roughness on fatigue strength prediction. Two additional terms were added to the standard Murakami formula: one to account for the size effect and the other to consider the nonzero value of the average stress. A cylindrical plain and a miniaturized plain specimens, in the as-built condition, were employed to calibrate the proposed model, while a machined plain cylindrical and a miniaturized notched specimens were used as validators with low prediction errors.

Near-infrared reflectance spectroscopy for rapid prediction of biochemical methane potential of wastewater wasted sludge

The information of biochemical methane potential (BMP) of wasted sludge is essential to ensure the stable operation of sludge management processes. However, conventional anaerobic digestion (AD) approach for BMP test is time-consuming and labour-intensive. Currently, the technique of Near Infrared Spectroscopy (NIRS) is gaining prominence in the biogas production within AD process. Previous studies mostly focused on predicting BMP values for fibrous plant biomass and solid waste, with only a limited number of studies attempting to apply NIRS to obtain BMP values across a wide array of wasted sludge types. To obtain BMP values for this diverse range of wasted sludge efficiently and accurately, it is imperative to develop precise models for assessing BMP values using NIRS. In this study, the possibility of using NIRS to predict the BMP values of wasted sludge was evaluated. A total of 70 sludge samples from different sources were investigated to develop a BMP-prediction model by correlating the measured BMP values with the obtained NIR spectra. As a result, a reliable and successful BMP-prediction model was established with the determination coefficient of 0.90, residual prediction deviation of 3.50 and low root mean square error of prediction of 36.8 mL CH4/g VS. This BMP-prediction model is satisfactory for predicting BMP values of various types of sludge. It could provide support for plant operators to make decisions rapidly, thereby improving the process efficiency and optimizing sludge management procedures.

Trajectory of reward-related abnormalities in unaffected relatives of patients with bipolar disorder – A longitudinal fMRI study

First-degree relatives of patients with bipolar disorder are at heightened risk of mood episodes, which may be attributed to the existence of endophenotypes i.e., heritable (neuro)biological changes present in patients and their unaffected relatives (UR). In this longitudinal MRI study, we aim to investigate the trajectories of aberrant reward-related functional changes identified in UR vs healthy controls (HC). Sixty-eight UR and 65 HC of similar age and gender distribution underwent MRI at baseline while performing a card guessing task. Of these, 29 UR and 36 HC were investigated with the same protocol following a 16-month period in average. We first identified brain regions showing group differences in the neural response to expected value (EV) and reward prediction error (PE) at baseline and analyzed how the reward-related response in these regions changed over time in UR vs HC. Relative to HC at baseline, UR showed lower EV signal in the right ventrolateral prefrontal cortex (vlPFC) and paracingulate gyrus and lower PE signal in the left vlPFC and dorsomedial PFC. The trajectories of these abnormalities in UR showed a normalization of the prefrontal EV signals, whereas the PE signals which correlated with depressive symptoms remained stable over time. While the UR showed both blunted EV and PE signals, none of these abnormalities increased over time, which is consistent with the observed stable mood symptoms.

Deep learning based denoising of super-resolution images of protein aggregates

Neurodegenerative diseases (NDs) are closely associated with the amyloid aggregation of proteins like Amyloid-, -synuclein, and tau. Understanding the pathogenesis of NDs requires studying the structures and morphological features of these aggregates, which are typically below 100 nm in size. Traditional fluorescence microscopy is limited by the diffraction limit of light (~250 nm). Single-molecule localization microscopy (SMLM) offers a resolution down to ~20 nm, enabling the visualization of these aggregates. However, issues such as non-specific binding (NSB) of fluorophores and background noise degrade the quality of SMLM images. This study presents a U-net-based convolutional neural network (CNN) to denoise SMLM images of protein aggregates. The training dataset includes noise-free super-resolution images of aggregates and their noisy counterparts with non-specific binding signals. Various imaging conditions are simulated to mimic real-world scenarios. The U-net's output is evaluated against ground-truth images for denoising performance. Post-processing techniques further enhance denoised images. The fine-tuned U-net model achieves a validation loss of 0.0042 and low prediction errors of 0.32% and 3.71% in the area and number of aggregates, respectively. This research offers a powerful tool for denoising SMLM images, facilitating accurate characterization of protein aggregate structures and morphological features.

The Enhanced Wagner–Hagras OLS–BP Hybrid Algorithm for Training IT3 NSFLS-1 for Temperature Prediction in HSM Processes

This paper presents (a) a novel hybrid learning method to train interval type-1 non-singleton type-3 fuzzy logic systems (IT3 NSFLS-1), (b) a novel method, named enhanced Wagner–Hagras (EWH) applied to IT3 NSFLS-1 fuzzy systems, which includes the level alpha 0 output to calculate the output y alpha using the average of the outputs y alpha k instead of their weighted average, and (c) the novel application of the proposed methodology to solve the problem of transfer bar surface temperature prediction in a hot strip mill. The development of the proposed methodology uses the orthogonal least square (OLS) method to train the consequent parameters and the backpropagation (BP) method to train the antecedent parameters. This methodology dynamically changes the parameters of only the level alpha 0, minimizing some criterion functions as new information becomes available to each level alpha k. The precursor sets are type-2 fuzzy sets, the consequent sets are fuzzy centroids, the inputs are type-1 non-singleton fuzzy numbers with uncertain standard deviations, and the secondary membership functions are modeled as two Gaussians with uncertain standard deviation and the same mean. Based on the firing set of the level alpha 0, the proposed methodology calculates each firing set of each level alpha k to dynamically construct and update the proposed EWH IT3 NSFLS-1 (OLS–BP) system. The proposed enhanced fuzzy system and the proposed hybrid learning algorithm were applied in a hot strip mill facility to predict the transfer bar surface temperature at the finishing mill entry zone using, as inputs, (1) the surface temperature measured by the pyrometer located at the roughing mill exit and (2) the time taken to translate the transfer bar from the exit of the roughing mill to the entry of the descale breaker of the finishing mill. Several fuzzy tools were used to make the benchmarking compositions: type-1 singleton fuzzy logic systems (T1 SFLS), type-1 adaptive network fuzzy inference systems (T1 ANFIS), type-1 radial basis function neural networks (T1 RBFNN), interval singleton type-2 fuzzy logic systems (IT2 SFLS), interval type-1 non-singleton type-2 fuzzy logic systems (IT2 NSFLS-1), type-2 ANFIS (IT2 ANFIS), IT2 RBFNN, general singleton type-2 fuzzy logic systems (GT2 SFLS), general type-1 non-singleton type-2 fuzzy logic systems (GT2 NSFLS-1), interval singleton type-3 fuzzy logic systems (IT3 SFLS), and interval type-1 non-singleton type-3 fuzzy systems (IT3 NSFLS-1). The experiments show that the proposed EWH IT3 NSFLS-1 (OLS–BP) system presented superior capability to learn the knowledge and to predict the surface temperature with the lower prediction error.

Measuring the Performance of Supervised Machine Learning Algorithms for Optimizing Wheat Productivity Prediction Models: A Comparative Study

The issue of precise crop prediction gained worldwide attention in the midst of food security concerns. In this study, the efficacies of different machine learning (ML) algorithms, i.e., multiple linear regression (MLR), decision tree regression (DTR), random forest regression (RFR), and support vector regression (SVR) are integrated to predict wheat productivity. The performances of ML algorithms are then measured to get the optimized model. The updated dataset is collected from the Crop Reporting Service for various agronomical constraints. Randomized data partitions, hyper-parametric tuning, complexity analysis, cross-validation measures, learning curves, evaluation metrics and prediction errors are used to get the optimized model. ML model is applied using 75% training dataset and 25% testing datasets. RFR achieved the highest R2 value of 0.90 for the training model, followed by DTR, MLR, and SVR. In the testing model, RFR also achieved an R2 value of 0.74, followed by MLR, DTR, and SVR. The lowest prediction error (P.E) is found for the RFR, followed by DTR, MLR, and SVR. K-Fold cross-validation measures also depict that RFR is an optimized model when compared with DTR, MLR and SVR.

Neural Network Estimators for Optimal Tour Lengths of Traveling Salesperson Problem Instances with Arbitrary Node Distributions

It is essential to solve complex routing problems to achieve operational efficiency in logistics. However, because of their complexity, these problems are often tackled sequentially using cluster-first, route-second frameworks. Unfortunately, such two-phase frameworks can suffer from suboptimality due to the initial phase. To address this issue, we propose leveraging information about the optimal tour lengths of potential clusters as a preliminary step, transforming the two-phase approach into a less myopic solution framework. We introduce quick and highly accurate Traveling Salesperson Problem (TSP) tour length estimators based on neural networks (NNs) to facilitate this. Our approach combines the power of NNs and theoretical knowledge in the routing domain, utilizing a novel feature set that includes node-level, instance-level, and solution-level features. This hybridization of data and domain knowledge allows us to achieve predictions with an average deviation of less than 0.7% from optimality. Unlike previous studies, we design and employ new instances replicating real-life logistics networks and morphologies. These instances possess characteristics that introduce significant computational costs, making them more challenging. To address these challenges, we develop a novel and efficient method for obtaining lower bounds and partial solutions to the TSP, which are subsequently utilized as solution-level predictors. Our computational study demonstrates a prediction error up to six times lower than the best machine learning (ML) methods on their training instances and up to 100 times lower prediction error on out-of-distribution test instances. Furthermore, we integrate our proposed ML models with metaheuristics to create an enumeration-like solution framework, enabling the improved solution of massive-scale routing problems. In terms of solution time and quality, our approach significantly outperforms the state-of-the-art solver, demonstrating the potential of our features, models, and the proposed method. History: This paper has been accepted for the Transportation Science Special Issue on Machine Learning Methods and Applications in Large-Scale Route Planning Problems. Supplemental Material: The online appendix is available at https://doi.org/10.1287/trsc.2022.0015 .

Transcranial magnetic stimulation mapping of the motor cortex: comparison of five estimation algorithms.

There are currently five different kinds of transcranial magnetic stimulation (TMS) motor mapping algorithms available, from ordinary point-based algorithms to advanced field-based algorithms. However, there have been only a limited number of comparison studies conducted, and they have not yet examined all of the currently available algorithms. This deficiency impedes the judicious selection of algorithms for application in both clinical and basic neuroscience, and hinders the potential promotion of a potential superior algorithm. Considering the influence of algorithm complexity, further investigation is needed to examine the differences between fMRI peaks and TMS cortical hotspots that were identified previously. Twelve healthy participants underwent TMS motor mapping and a finger-tapping task during fMRI. The motor cortex TMS mapping results were estimated by five algorithms, and fMRI activation results were obtained. For each algorithm, the prediction error was defined as the distance between the measured scalp hotspot and optimized coil position, which was determined by the maximum electric field strength in the estimated motor cortex. Additionally, the study identified the minimum number of stimuli required for stable mapping. Finally, the location difference between the TMS mapping cortical hotspot and the fMRI activation peak was analyzed. The projection yielded the lowest prediction error (5.27 ± 4.24 mm) among the point-based algorithms and the association algorithm yielded the lowest (6.66 ± 3.48 mm) among field-based estimation algorithms. The projection algorithm required fewer stimuli, possibly resulting from its suitability for the grid-based mapping data collection method. The TMS cortical hotspots from all algorithms consistently deviated from the fMRI activation peak (20.52 ± 8.46 mm for five algorithms). The association algorithm might be a superior choice for clinical applications and basic neuroscience research, due to its lower prediction error and higher estimation sensitivity in the deep cortical structure, especially for the sulcus. It also has potential applicability in various other TMS domains, including language area mapping and more. Otherwise, our results provide further evidence that TMS motor mapping intrinsically differs from fMRI motor mapping.

Ensemble Quantitative Read-Across Structure-Activity Relationship Algorithm for Predicting Skin Cytotoxicity.

Read-across (RA) and quantitative structure-activity relationship (QSAR) are two alternative methods commonly used to fill data gaps in chemical registrations. These approaches use physicochemical properties or molecular fingerprints of source substances to predict the properties of unknown substances that have similar chemical structures or physicochemical properties. Research on RA and QSAR is essential to minimize the time, money, and animal testing needed to determine biological properties that are not currently known. This study developed a stacked ensemble quantitative read-across structure-activity relationship algorithm (enQRASAR) for predicting skin irritation toxicity based on negative log cell viability inhibition concentration at 50% (pIC50) against skin keratinocytes as the end point. The goodness-of-fit and predictability of this algorithm were validated using leave-one-out cross-validation and external test data sets. The results obtained were statistically reliable in terms of goodness-of-fit, robustness, and predictability metrics. Additionally, the developed model demonstrated a low prediction error when predicting FDA-approved drugs. These results confirm that the enQRASAR algorithm can be used to predict skin cytotoxicity of chemicals. Therefore, this model was publicly available to further facilitate toxicity predictions of unknown compounds in chemical registrations.

Enhanced abnormal data detection hybrid strategy based on heuristic and stochastic approaches for efficient patients rehabilitation

Over the last few years, substantial research has been conducted towards developing efficient abnormal detection techniques while considering efficiency, accuracy, high-dimensional data, distributed environments, and others. Researchers increasingly deal with “abnormalities” in clinical patient data to derive relevant clinical knowledge for making informed decisions. However, data collection for clinically relevant research is often guided by patient conditions and administrative or clinical requirements rather than a regular schedule. Therefore, clinical data is frequently obtained in an unreliable form, characterized by data outliers and inconsistencies, incomplete information, and an unstructured format that varies based on patient types and data structures. In this research study, an enhanced hybrid AD strategy is developed based on heuristic and stochastic methods to cope with abnormalities in the clinical data of patients. The proposed hybrid strategy employs optimal k-means clustering as a heuristic method to cluster the clinical data based on the patient’s routine exercise characteristics to cope with abnormalities efficiently. Next, an interquartile range-based stochastic approach is employed as a statistical method to detect and eliminate abnormal data points by providing only reliable and effectual data to medical practitioners. The main objective of this research article is to facilitate healthcare and research practitioners by dealing with a high dimensional massive amount of inconsistent and incomplete clinical data of patients to detect and discard anomalous data points for providing only efficacious information. Furthermore, the AutoML paradigm is employed to develop an optimal regression model for analyzing the impact of the proposed hybrid strategy for abnormal pattern detection. In addition, different statistical error estimation measures are used to evaluate the empirical effectiveness of the proposed hybrid strategy using AutoML. The experiment results show a noteworthy improvement in terms of the R2 score for predicting healthcare indicators compared to the existing state-of-the-art regression models. Our optimal regression model performed efficiently regarding the R2 score and MAPE; it achieved an R2 score of 0.9855 and 0.9850 for predicting the Borg RPE and TUG, respectively. Similarly, our model achieved a low prediction error in terms of MAPE for predicting both health functional indicators; it achieved a MAPE of 6.57% and 5.19% for Borg RPE and TUG prediction. Our contribution signifies that the performance of the AutoML improves and outperforms traditional regression models while applying our proposed hybrid abnormal detection model to the patient’s rehabilitation data for accurately dealing with anomalous data.

Prospective evaluation of the ESCRS online calculator for calculation of a multifocal intraocular lens.

To evaluate a recently introduced ESCRS online calculator for intraocular lens (IOL) calculation of a multifocal IOL in refractive lens exchange and cataract surgery in a prospective setting. Department of Ophthalmology, Goethe University Frankfurt, Germany. Prospective, consecutive case series. We included eyes that received lens extraction and multifocal IOL implantation. We compared the mean prediction error (MPE), mean and median absolute prediction error (MAE, MedAE) of the online calculator provided by the European Society of Cataract and Refractive Surgeons (ESCRS), and number of eyes within ±0.5, ±1.0, ±2.0 diopters (D) of target refraction. Additionally, the SRK/T formula was included for comparison as well. Postoperative spherical equivalent was measured at 3 months. Only one eye per patient was included. Eighty-eight eyes from 88 patients with mean age of 62 years ± 9.5 were included. The MedAE was low for all formulas and ranged from 0.26D (Kane), Hill-RBF (0.27D), Hoffer QST (0.27D), Barrett (0.28D), EVO (0.29D), Cooke K6 (0.27D), 0.30D (Pearl DGS) to 0.31D (SRK/T). No statistically significant difference was found (p = 0.627). Considering the number of eyes within ±0.5D of the calculated refraction the best performing was again the Hill-RBF (84%, 74 eyes), again followed by Kane (71, 81%), EVO, Pearl DGS, Hoffer QST, Barrett (each 80%, 70 eyes),Cooke K6 (78%, 69 eyes), and SRK/T (74%). Again, no statistically significant difference was found (p = 0.39). Using recently introduced ESCRS online IOL calculator in multifocal intraocular lenses leads to a high number of eyes reaching target refraction and low prediction errors. All formulas performed similar well. However, Hill-RBF showed the highest number of eyes within ±0.5D, but no significance was found regarding this.

Data‐driven train delay prediction incorporating dispatching commands: An XGBoost‐metaheuristic framework

AbstractTrain delays can significantly impact the punctuality and service quality of high‐speed trains, which also play a crucial role in affecting dispatchers with their decision‐making. In this study, a data‐driven train delay prediction framework was proposed and strengthened by considering the impact of dispatching commands and the mechanisms of train delay propagation using XGBoost. Four metaheuristic algorithms were utilized to fine‐tune its hyperparameters. A vast dataset comprising 1.9 million records spanning 38 months of train operation data was utilized for feature extraction and model training. The model's accuracy was evaluated using three statistical metrics, and a comparison of the four tuning frameworks was performed. To emphasize the model's interpretability and its practical guidance for train rescheduling, the relationship of dispatching commands, delay propagation and delay prediction was validated by combining the theory and practical results, and a SHAP (SHapley Additive exPlanations) analysis was used for a clearer model explanation. The results revealed that distinct XGBoost‐Metaheuristic models exhibit unique effects in different criteria, yet they all demonstrated high accuracy and low prediction errors, thereby revealing the potential of using machine learning for train delay prediction, which is valuable for decision‐making and rescheduling.

Resting energy expenditure prediction equations have low accuracy and high prediction error: From malnutrition to obesity

Resting energy expenditure prediction equations have low accuracy and high prediction error: From malnutrition to obesity

Prediction of Global Ionospheric Total Electron Content (TEC) Based on SAM‐ConvLSTM Model

AbstractThis paper first applies a prediction model based on self‐attention memory ConvLSTM (SAM‐ConvLSTM) to predict the global ionospheric total electron content (TEC) maps with up to 1 day of lead time. We choose the global ionospheric TEC maps released by the Center for Orbit Determination in Europe (CODE) as the training data set covering the period from 1999 to 2022. Besides that, we put several space environment data as additional multivariate‐features into the framework of the prediction model to enhance its forecasting ability. In order to confirm the efficiency of the proposed model, the other two prediction models based on convolutional long short‐term memory (LSTM) are used for comparison. The three models are trained and evaluated on the same data set. Results show that the proposed SAM‐ConvLSTM prediction model performs more accurately than the other two models, and more stably under space weather events. In order to assess the generalization capabilities of the proposed model amidst severe space weather occurrences, we selected the period of 22–25 April 2023, characterized by a potent geomagnetic storm, for experimental validation. Subsequently, we employed the 1‐day predicted global TEC products from the Center for Operational Products and Services (COPG) and the SAM‐ConvLSTM model to evaluate their respective forecasting prowess. The results show that the SAM‐ConvLSTM prediction model achieves lower prediction error. In one word, the ionospheric TEC prediction model proposed in this paper can establish the ionosphere TEC of spatio‐temporal data association for a long time, and realize high precision of prediction performance.

Improving a regional peat thickness map using soil apparent electrical conductivity measurements at the field-scale

IntroductionThe increased adoption of proximal sensors has helped to generate peat mapping products: they gather data quickly and can detect the peat-mineral later boundary. A third layer, made of sedimentary peat (limnic layers, gyttja), can sometimes be found in between them. This material is highly variable spatially and is associated with degraded soil properties when located near the surface.MethodsThis study aimed to assess the potential of direct current resistivity measurements to predict the maximum peat thickness (MPT), defined as the non-limnic peat thickness, to facilitate soil conservation and management practices at the field-scale. The results were also compared to a regional map of the MPT from a previous study used and also tested as a covariate. This study was conducted in a shallow (MPT = 8-138 cm) cultivated organic soil from Québec, Canada. The MPT was mapped using the apparent electrical conductivity (ECa) from a Veris Q2800, and a digital elevation model, with and without a regional MPT map (RM) as a covariate to downscale it. Three machine-learning algorithms (Cubist, Random Forest, and Support Vector Regression) were compared to ordinary kriging (OK), multiple linear regression, and multiple linear regression kriging (MLRK) models.Results and discussionThe best predictive performance was achieved with OK (Lin’s CCC = 0.89, RMSE = 13.75 cm), followed by MLRK-RM (CCC = 0.85, RMSE = 15.7 cm). All models were more accurate than the RM (CCC = 0.65, RMSE = 29.85 cm), although they underpredicted MPT &amp;gt; 100 cm. Moreover, the addition of the RM as a covariate led to a lower prediction error and higher accuracy for all models. Overall, a field-scale approach could better support precision soil conservation interventions by generating more accurate management zones. Future studies should test multi-sensor fusion and other geophysical sensors to further improve the model performance and detect deeper boundaries.

Computation Offloading for Energy and Delay Trade-Offs With Traffic Flow Prediction in Edge Computing-Enabled IoV

An unprecedented prosperity in artificial intelligence promotes the development of Internet of Vehicles (IoV). Assisted by edge computing, vehicles enable to offload data to edge servers in close proximity to users for processing, thus making up for the shortage of local computing resources. However, due to the uneven space-time distribution of traffic flow, edge servers of a certain road segment may be overwhelmed by the surge of service requests. Furthermore, IoV system will incur significant additional energy consumption and time delay because of the absence of a proper computation offloading scheme between edge servers. To cope with above challenges, a computing offloading method for energy and delay trade-offs with traffic flow prediction in edge computing-enabled IoV is proposed. We first design the graph weighted convolution network (GWCN) that can fully excavate the connectivity and distance relation information between road segments to conduct traffic flow prediction. The short-term prediction results are utilized as the basis for adjusting the resource allocation of edge resources in different regions. Then, a computation offloading method driven by deep deterministic policy gradient (DDPG) is leveraged to obtain an optimal computation offloading scheme for edge servers. Finally, extensive comparative experiments demonstrate the low prediction error of GWCN and superior performance of DDPG-driven method in reducing total time delay and energy consumption.