Similar Predictive Accuracy Research Articles

Prediction of the soil organic carbon in the LUCAS soil database based on spectral clustering

The estimation of the level of the soil organic carbon (SOC) content plays an important role in assessing the soil health state. Visible and Near Infrared Diffuse Reflectance Spectroscopy (Vis-NIR DRS) is a fast and cheap tool for measuring the SOC. However, when this technology is applied on a larger area, the soil prediction accuracy decreases due to the heterogeneity of the samples. In this paper, we first investigate the global model performance in the LUCAS EU-wide topsoil database. Then, different clustering strategies were tested, including the k-means clustering based on the principal component analysis (PCA) and hierarchical clustering, combined with the partial least squares regression (PLSR) models, and a clustering based on a local PLSR approach. The best validation results were obtained for the local PLSR approach with R2 = 0.75, root mean squared error of prediction (RMSEP) = 13.38 g/kg and ratio of performance to interquartile range (RPIQ) = 2.846, but the algorithm running time was 30.05 s. Similar results were obtained for the k-means clustering method with R2 = 0.75, RMSEP = 14.61 g/kg and RPIQ = 2.844, at only 4.52 s. This study demonstrates that the PLSR approach based on k-means clustering is able to achieve similar prediction accuracy as the local PLSR approach, while significantly improving the algorithm speed. This provides the theoretical basis for adapting the spectral soil model to the needs of real-time SOC quantification.

An online autonomous learning and prediction scheme for machine learning assisted structural optimization

In this paper, an online autonomous learning and prediction (ALP) scheme is proposed for machine learning assisted structural optimization (MLaSO) to reduce total computational time with controlled accuracy of objective function prediction. In MLaSO, information in selected routine iterations is used via online learning to train a Neural Network, which is then used to perform online prediction of information in subsequent other iterations. Although using MLaSO with pre-defined learning and prediction scheme can save total computational time, it is difficult to determine the most suitable learning and prediction scheme prior to solving an optimization problem. The present ALP method enables autonomous activation of online learnings and predictions via maximizing specific computational efficiency while constraining the accuracy of the objective function prediction. Four numerical 2D and 3D examples are presented to demonstrate the computational benefits of integrating ALP with gradient based MLaSO method. The present numerical results demonstrate that: (a) using ALP in MLaSO can save more computational time than using the pre-defined scheme with similar prediction accuracy, and (b) more computational time can be saved by allowing larger tolerance in the accuracy of objective function prediction.

Interpreting Infrared Thermography with Deep Learning to Assess the Mortality Risk of Critically Ill Patients at Risk of Hypoperfusion.

Hypoperfusion, a common manifestation of many critical illnesses, could lead to abnormalities in body surface thermal distribution. However, the interpretation of thermal images is difficult. Our aim was to assess the mortality risk of critically ill patients at risk of hypoperfusion in a prospective cohort by infrared thermography combined with deep learning methods. This post-hoc study was based on a cohort at high-risk of hypoperfusion. Patients' legs were selected as the region of interest. Thermal images and conventional hypoperfusion parameters were collected. Six deep learning models were attempted to derive the risk of mortality (range: 0 to 100%) for each patient. The area under the receiver operating characteristic curve (AUROC) was used to evaluate predictive accuracy. Fifty-five hospital deaths occurred in a cohort consisting of 373 patients. The conventional hypoperfusion (capillary refill time and diastolic blood pressure) and thermal (low temperature area rate and standard deviation) parameters demonstrated similar predictive accuracies for hospital mortality (AUROC 0.73 and 0.77). The deep learning methods, especially the ResNet (18), could further improve the accuracy. The AUROC of ResNet (18) was 0.94 with a sensitivity of 84% and a specificity of 91% when using a cutoff of 36%. ResNet (18) presented a significantly increasing trend in the risk of mortality in patients with normotension (13 [7 to 26]), hypotension (18 [8 to 32]) and shock (28 [14 to 62]). Interpreting infrared thermography with deep learning enables accurate and non-invasive assessment of the severity of patients at risk of hypoperfusion.

Forecasting a Commodity-Exporting Small Open Developing Economy Using DSGE and DSGE-BVAR

In this paper, we assess the forecasting performance of three types of structural models – DSGE, BVAR with Minnesota priors, and DSGE-BVAR – in the context of a commodity-exporting small open developing economy using the data for Kazakhstan. We find that BVAR and DSGE-BVAR models generally produce point forecasts that are more accurate and less biased compared to those of DSGE in the short term, but that BVAR forecasts rapidly deteriorate in quality as the length of the forecast horizon increases. The density forecast analysis shows that when all variables are jointly considered, the models have similar prediction accuracy, and when financial sector variables are omitted, the BVAR and DSGE-BVAR models demonstrate superior performance in the short term.

Comparing Individualized Survival Predictions From Random Survival Forests and Multistate Models in the Presence of Missing Data: A Case Study of Patients With Oropharyngeal Cancer.

In recent years, interest in prognostic calculators for predicting patient health outcomes has grown with the popularity of personalized medicine. These calculators, which can inform treatment decisions, employ many different methods, each of which has advantages and disadvantages. We present a comparison of a multistate model (MSM) and a random survival forest (RSF) through a case study of prognostic predictions for patients with oropharyngeal squamous cell carcinoma. The MSM is highly structured and takes into account some aspects of the clinical context and knowledge about oropharyngeal cancer, while the RSF can be thought of as a black-box non-parametric approach. Key in this comparison are the high rate of missing values within these data and the different approaches used by the MSM and RSF to handle missingness. We compare the accuracy (discrimination and calibration) of survival probabilities predicted by both approaches and use simulation studies to better understand how predictive accuracy is influenced by the approach to (1) handling missing data and (2) modeling structural/disease progression information present in the data. We conclude that both approaches have similar predictive accuracy, with a slight advantage going to the MSM. Although the MSM shows slightly better predictive ability than the RSF, consideration of other differences are key when selecting the best approach for addressing a specific research question. These key differences include the methods' ability to incorporate domain knowledge, and their ability to handle missing data as well as their interpretability, and ease of implementation. Ultimately, selecting the statistical method that has the most potential to aid in clinical decisions requires thoughtful consideration of the specific goals.

Predictive discriminative accuracy of walking abilities at discharge for community ambulation levels at 6 months post-discharge among inpatients with subacute stroke.

[Purpose] This study aimed to compare the predictive accuracy of walking ability at discharge among subacute stroke inpatients at 6 months post-discharge in terms of community ambulation level and establish optimal cut-off values. [Participants and Methods] This prospective observational study included 78 patients who completed follow-up assessments. Patients were classified into three groups based on the Modified Functional Walking Category (household/most limited community walkers, least limited community walkers, and unlimited community walkers) obtained by telephone survey at 6 months post-discharge. Predictive accuracy and cut-off values for discriminating among groups were calculated from 6-minute walking distance and comfortable walking speed measured at the time of discharge using receiver operating characteristic curves. [Results] Between household/most limited and least limited community walkers, 6-minute walking distance and comfortable walking speed offered similar predictive accuracy (area under the curve, 0.6-0.7), with cut-off values of 195 m and 0.56 m/s, respectively. Between least limited and unlimited community walkers, the areas under the curve were 0.896 for 6-minute walking distance and 0.844 for comfortable walking speed, with cut-off values of 299 m and 0.94 m/s, respectively. [Conclusion] Walking endurance and walking speed among inpatients with subacute stroke provided superior predictive accuracy for unlimited community walkers at 6 months post-discharge.

Relationship between Neutrophil-to-HDL Cholesterol Ratio and Mortality in Hospitalized HBV-Associated Decompensated Cirrhosis.

Early prognosis prediction in HBV-associated decompensated cirrhosis (HBV-DC) is important to decrease its high mortality. This study aimed to determine the relationship between neutrophil-to-high-density lipoprotein cholesterol ratio (NHR) and 30-day survival in HBV-DC patients. This retrospective study involved HBV-DC patients in our hospital from September 2020 to December 2022. The main outcome was 30-day survival. A multivariate analysis was performed to determine whether NHR influenced 30-day survival in the patients, and receiver operating characteristic (ROC) analyses were conducted to calculate the prognostic accuracy of NHR versus Model for End-Stage Liver Disease (MELD) score. In this study, 146 HBV-DC patients were included, and 23 cases (15.8%) died within 30 days. The NHR values differed markedly between non-survivors and survivors, and high NHR was associated with an increased risk of adverse outcomes. On ROC analyses, NHR showed similar predictive accuracy as MELDs for predicting mortality in HBV-DC. The present study suggests that NHR could be a useful new prognostic tool in HBV-DC patients.

Natural language processing and machine learning approaches for food categorization and nutrition quality prediction compared with traditional methods

BackgroundFood categorization and nutrient profiling are labor intensive, time consuming, and costly tasks, given the number of products and labels in large food composition databases and the dynamic food supply.Objectives: This study used a pretrained language model and supervised machine learning to automate food category classification and nutrition quality score prediction based on manually coded and validated data, and compared prediction results with models using bag-of-words and structured nutrition facts as inputs for predictions. MethodsFood product information from University of Toronto Food Label Information and Price Database 2017 (n = 17,448) and University of Toronto Food Label Information and Price Database 2020 (n = 74,445) databases were used. Health Canada’s Table of Reference Amounts (TRA) (24 categories and 172 subcategories) was used for food categorization and the Food Standards of Australia and New Zealand (FSANZ) nutrient profiling system was used for nutrition quality score evaluation. TRA categories and FSANZ scores were manually coded and validated by trained nutrition researchers. A modified pretrained sentence-Bidirectional Encoder Representations from Transformers model was used to encode unstructured text from food labels into lower-dimensional vector representations, followed by supervised machine learning algorithms (i.e., elastic net, k-Nearest Neighbors, and XGBoost) for multiclass classification and regression tasks. ResultsPretrained language model representations utilized by the XGBoost multiclass classification algorithm reached overall accuracy scores of 0.98 and 0.96 in predicting food TRA major and subcategories, outperforming bag-of-words methods. For FSANZ score prediction, our proposed method reached a similar prediction accuracy (R2: 0.87 and MSE: 14.4) compared with bag-of-words methods (R2: 0.72–0.84; MSE: 30.3–17.6), whereas structured nutrition facts machine learning model performed the best (R2: 0.98; MSE: 2.5). The pretrained language model had a higher generalizable ability on the external test datasets than bag-of-words methods. ConclusionsOur automation achieved high accuracy in classifying food categories and predicting nutrition quality scores using text information found on food labels. This approach is effective and generalizable in a dynamic food environment, where large amounts of food label data can be obtained from websites.

An efficient genomic prediction method without the direct inverse of the genomic relationship matrix.

GBLUP, the most widely used genomic prediction (GP) method, consumes large and increasing amounts of computational resources as the training population size increases due to the inverse of the genomic relationship matrix (GRM). Therefore, in this study, we developed a new genomic prediction method (RHEPCG) that avoids the direct inverse of the GRM by combining randomized Haseman-Elston (HE) regression (RHE-reg) and a preconditioned conjugate gradient (PCG). The simulation results demonstrate that RHEPCG, in most cases, not only achieves similar predictive accuracy with GBLUP but also significantly reduces computational time. As for the real data, RHEPCG shows similar or better predictive accuracy for seven traits of the Arabidopsis thaliana F2 population and four traits of the Sorghum bicolor RIL population compared with GBLUP. This indicates that RHEPCG is a practical alternative to GBLUP and has better computational efficiency.

On the use of machine learning methods to improve the estimation of gross primary productivity of maize field with drip irrigation

As an important part of the energy-matter cycle, gross primary productivity (GPP) reflects the ability of plants to absorb carbon dioxide from the atmosphere. The accurate estimation of GPP is critical to understanding the regional carbon cycle. With the development of machine learning (ML) theory, machine learning models are increasingly used to study complex phenomena with high variability in time and space. In this study, three machine learning models (Support vector regression (SVR), artificial neural network (ANN), and long short-term memory networks (LSTM)) were investigated for predicting GPP in northwest China and compared them with the traditional physical models. Carbon flux, various environmental factors, and maize growth indices were measured in the maize field over five years in northwest China. The rigorous analysis which included statistical comparison and cross-validation for the prediction of GPP confirmed that the machine learning models performed better than traditional physical models. And the SVR model performed the best among the considered ML models with the highest nash-efficiency coefficient and the lowest root mean squared error. The machine learning model also outperformed the traditional physical models on cloudy days and after irrigation. The SVR achieved good prediction accuracy and high stability. With different training data sets, the ANN and LSTM were relatively more sensitive to the training data set. When the training data was sufficient, SVR, ANN and LSTM could achieve similar prediction accuracy, but SVR was slightly higher. When the training data was small, the simulation accuracy of SVR was better than ANN and LSTM. The performance of ANN and LSTM was more sensitive to parameter selection, and the relationship between model performance and parameter selection had no obvious regularity. Based on this comprehensive comparison study, it was elicited that the SVR model can be successfully applied to GPP simulation of maize fields, which provided a new perspective for the application of machine learning modeling in GPP simulation.

Investigating ‘greyness’ of hybrid model for bioprocess predictive modelling

Hybrid modelling combines data-driven and mechanistic modelling, providing a cost-effective solution to modelling complex biochemical reaction kinetics when the underlying mechanisms are not fully understood. However, the question of how much kinetic information to incorporate into a hybrid bioprocess model has not been systematically addressed. Therefore, this work built three hybrid models for predicting the temperature-dependent rates of biomass growth, glucose consumption and γ-linolenic acid accumulation during fermentation of the fungus Cunninghamella echinulata. Each hybrid model incorporated different amounts of kinetic information from a pre-existing complex kinetic model, representing three levels of hybrid model ‘greyness’, then embedded a Gaussian Process (GP) to simulate the unknown kinetics inferred from experimental measurements. Although all three hybrid models fitted well, incorporating more specific kinetic information increased the risk of incorporating incorrect inductive bias that hindered rather than enhanced hybrid model performance. This observation also held when using the hybrid models to predict the temperature-shift dynamics for the upscaled 5 L bioreactor. Nonetheless, the hybrid models demonstrated much improved predictive confidence with similar predictive accuracy to the original kinetic model, demonstrating the proficiency of hybrid modelling for accelerating the construction of confident bioprocess models for robust process optimisation and real-time monitoring.

SNP- and haplotype-based single-step genomic predictions for body weight, wool, and reproductive traits in North American Rambouillet sheep.

Rambouillet sheep are commonly raised in extensive grazing systems in the US, mainly for wool and meat production. Genomic evaluations in US sheep breeds, including Rambouillet, are still incipient. Therefore, we aimed to evaluate the feasibility of performing genomic prediction of breeding values for various traits in Rambouillet sheep based on single nucleotide polymorphisms (SNP) or haplotypes (fitted as pseudo-SNP) under a single-step GBLUP approach. A total of 28,834 records for birth weight (BWT), 23,306 for postweaning weight (PWT), 5,832 for yearling weight (YWT), 9,880 for yearling fibre diameter (YFD), 11,872 for yearling greasy fleece weight (YGFW), and 15,984 for number of lambs born (NLB) were used in this study. Seven hundred forty-one individuals were genotyped using a moderate (50 K; n=677) or high (600 K; n=64) density SNP panel, in which 32 K SNP in common between the two SNP panels (after genotypic quality control) were used for further analyses. Single-step genomic predictions using SNP (H-BLUP) or haplotypes (HAP-BLUP) from blocks with different linkage disequilibrium (LD) thresholds (0.15, 0.35, 0.50, 0.65, and 0.80) were evaluated. We also considered different blending parameters when constructing the genomic relationship matrix used to predict the genomic-enhanced estimated breeding values (GEBV), with alpha equal to 0.95 or 0.50. The GEBV were compared to the estimated breeding values (EBV) obtained from traditional pedigree-based evaluations (A-BLUP). The mean theoretical accuracy ranged from 0.499 (A-BLUP for PWT) to 0.795 (HAP-BLUP using haplotypes from blocks with LD threshold of 0.35 and alpha equal to 0.95 for YFD). The prediction accuracies ranged from 0.143 (A-BLUP for PWT) to 0.330 (A-BLUP for YGFW) while the prediction bias ranged from -0.104 (H-BLUP for PWT) to 0.087 (HAP-BLUP using haplotypes from blocks with LD threshold of 0.15 and alpha equal to 0.95 for YGFW). The GEBV dispersion ranged from 0.428 (A-BLUP for PWT) to 1.035 (A-BLUP for YGFW). Similar results were observed for H-BLUP or HAP-BLUP, independently of the LD threshold to create the haplotypes, alpha value, or trait analysed. Using genomic information (fitting individual SNP or haplotypes) provided similar or higher prediction and theoretical accuracies and reduced the dispersion of the GEBV for body weight, wool, and reproductive traits in Rambouillet sheep. However, there were no clear improvements in the prediction bias when compared to pedigree-based predictions. The next step will be to enlarge the training populations for this breed to increase the benefits of genomic predictions.

Satellite prediction of coastal hypoxia in the northern Gulf of Mexico

The growing number, size, and frequency of coastal hypoxia increasingly threaten marine ecosystem health and essential ecosystem services for human well-being. It is therefore urgent to use continuous and consistent observation and develop advanced tools to characterize and track the spatial and temporal change of coastal hypoxia. Satellite imagery with fine spatiotemporal resolution and global coverage has shown great potential for monitoring environmental changes, yet has rarely been applied to hypoxia mapping. To advance the understanding, we synthesized satellite-derived ocean color variables and dissolved oxygen measurements collected during 2014, and used random forest regression, lagged linear regression, and functional data analysis to estimate the spatiotemporal change of the hypoxia zone in the Gulf of Mexico. The three models achieved similar predictive accuracy (±1.2–1.4 mg/L dissolved oxygen), but the random forest regression performed the best in estimating the bottom dissolved oxygen from satellite-derived variables. Our models also revealed time lags of roughly 0–5 and 16–19 days between the surface water process (e.g., algae bloom and ocean warming) and bottom water hypoxia, which was rarely considered in previous hypoxia studies using satellite data. Finally, our models showed that the area of Gulf hypoxia increased gradually from May and reached a peak during mid-July and mid-August in 2014, and the hypoxia zone occurred in the estuary of the Mississippi River and Suwannee River during roughly 25% of summer days. In addition to predicting the size of hypoxic zones, our study provides additional information on where, when, and how long hypoxic zones persist with greater spatial details and enables modeling hypoxic zones at near-real-time (e.g., days) temporal scales. More importantly, we demonstrate the great potential of applying satellite remote sensing for spatially explicit hypoxia mapping, which could promote more cost-effective coastal hypoxia monitoring and assessment practices.

Predicting Functional Outcome Using 24-Hour Post-Treatment Characteristics: Application of Machine Learning Algorithms in the STRATIS Registry.

ML models, adaptive boost, random forest (RF), classification and regression trees (CART), C5.0 decision tree (C5.0), support vector machine (SVM), least absolute shrinkage and selection operator (LASSO), and logistic regression (LR), and traditional LR models were used to predict 90-day functional outcome (modified Rankin Scale score 0-2). Twenty-four-hour National Institutes of Health Stroke Scale (NIHSS) was examined as a continuous or dichotomous variable in all models. Model accuracy was assessed using the area under characteristic curve (AUC). The 24-hour NIHSS score was a top-predictor of functional outcome in all models. ML models using the continuous 24-hour NIHSS scored showed moderate-to-good predictive performance (range mean AUC: 0.76-0.92); however, RF (AUC: 0.92 ± 0.028) outperformed all ML models, except LASSO (AUC: 0.89 ± 0.023, p= 0.0958). Importantly, RF demonstrated a significantly higher predictive value than LR (AUC: 0.87 ± 0.031, p= 0.048) and traditional LR (AUC: 85 ± 0.06, p= 0.035) when using the 24-hour continuous NIHSS score. Predictive accuracy was similar between the 24-hour NIHSS score dichotomous and continuous ML models. In this substudy, we found similar predictive accuracy for functional outcome when using the 24-hour NIHSS score as a continuous or dichotomous variable in ML models. ML models had moderate-to-good predictive accuracy, with RF outperforming LR models. External validation of these ML models is warranted. ANN NEUROL 2023;93:40-49.

GIFAIR-FL: A Framework for Group and Individual Fairness in Federated Learning

In this paper, we propose GIFAIR-FL, a framework that imposes group and individual fairness (GIFAIR) to federated learning (FL) settings. By adding a regularization term, our algorithm penalizes the spread in the loss of client groups to drive the optimizer to fair solutions. Our framework GIFAIR-FL can accommodate both global and personalized settings. Theoretically, we show convergence in nonconvex and strongly convex settings. Our convergence guarantees hold for both independent and identically distributed (i.i.d.) and non-i.i.d. data. To demonstrate the empirical performance of our algorithm, we apply our method to image classification and text prediction tasks. Compared with existing algorithms, our method shows improved fairness results while retaining superior or similar prediction accuracy. History: Kwok-Leung Tsui served as the senior editor for this article. Funding: This work was supported by NSF CAREER [Grant 2144147]. Data Ethics & Reproducibility Note: The code capsule is available on Code Ocean at https://codeocean.com/capsule/2590027/tree/v1 and in the e-Companion to this article (available at https://doi.org/10.1287/ijds.2022.0022 ).

Genetic analysis of hypermelanosis in Chinese tongue sole (Cynoglossus semilaevis)

Chinese tongue sole (Cynoglossus semilaevis) is an economically important marine fish in China. Generally, the eyeless side of the Chinese tongue sole is white and the side with eyes is brown after metamorphosis, hypermelanosis may still occur in the eyeless side in certain individuals after metamorphosis, which greatly decreases consumer acceptance and market price. In order to study the possibility of genetic improvement, we determined genomic markers in Chinese tongue sole using the genotyping-by-sequencing method and analyzed their association with hypermelanosis area. Genetic analysis showed that hypermelanosis was a complicated quantitative trait, and the estimated heritability for hypermelanosis incidence and area ratio were 0.16 and 0.21, respectively. Genomic selection analysis showed that selection based on hypermelanosis incidence and area ratio had similar reliabilities and prediction accuracies, indicating the feasibility of genetic improvement. Nine loci were significantly associated with hypermelanosis, few of which included genes or flanked genes potentially associated with skin disease, indicating the potential complicated genetic mechanisms underlying hypermelanosis in the Chinese tongue sole.

Impact of Atmospheric Correction Methods Parametrization on Soil Organic Carbon Estimation Based on Hyperion Hyperspectral Data

Visible Near infrared and Shortwave Infrared (VNIR/SWIR, 400–2500 nm) remote sensing data is becoming a tool for topsoil properties mapping, bringing spatial information for environmental modeling and land use management. These topsoil properties estimates are based on regression models, linking a key topsoil property to VNIR/SWIR reflectance data. Therefore, the regression model’s performances depend on the quality of both topsoil property analysis (measured on laboratory over-ground soil samples) and Bottom-of-Atmosphere (BOA) VNIR/SWIR reflectance which are retrieved from Top-Of-Atmosphere radiance using atmospheric correction (AC) methods. This paper examines the sensitivity of soil organic carbon (SOC) estimation to BOA images depending on two parameters used in AC methods: aerosol optical depth (AOD) in the FLAASH (Fast Line-of-Sight Atmospheric Analysis of Spectral Hypercubes) method and water vapor (WV) in the ATCOR (ATmospheric CORrection) method. This work was based on Earth Observing-1 Hyperion Hyperspectral data acquired over a cultivated area in Australia in 2006. Hyperion radiance data were converted to BOA reflectance using seven values of AOD (from 0.2 to 1.4) and six values of WV (from 0.4 to 5 cm), in FLAASH and ATCOR, respectively. Then a Partial Least Squares regression (PLSR) model was built from each Hyperion BOA data to estimate SOC over bare soil pixels. This study demonstrated that the PLSR models were insensitive to the AOD variation used in the FLAASH method, with R2cv and RMSEcv of 0.79 and 0.4%, respectively. The PLSR models were slightly sensitive to the WV variation used in the ATCOR method, with R2cv ranging from 0.72 to 0.79 and RMSEcv ranging from 0.41 to 0.47. Regardless of the AOD values, the PLSR model based on the best parametrization of the ATCOR model provided similar SOC prediction accuracy to PLSR models using the FLAASH method. Variation in AOD using the FLAASH method did not impact the identification of bare soil pixels coverage which corresponded to 82.35% of the study area, while a variation in WV using the ATCOR method provided a variation of bare soil pixels coverage from 75.04 to 84.04%. Therefore, this work recommends (1) the use of the FLAASH AC method to provide BOA reflectance values from Earth Observing-1 Hyperion Hyperspectral data before SOC mapping or (2) a careful selection of the WV parameter when using ATCOR.

Machine learning based survival prediction in Glioma using large-scale registry data.

Gliomas are the most common central nervous system tumors exhibiting poor clinical outcomes. The ability to estimate prognosis is crucial for both patients and providers in order to select the most appropriate treatment. Machine learning (ML) allows for sophisticated approaches to survival prediction using real world clinical parameters needed to achieve superior predictive accuracy. We employed Cox Proportional hazards (CPH) model, Support Vector Machine (SVM) model, Random Forest (RF) model in a large glioma dataset (3462 patients, diagnosed 2000-2018) to explore the most optimal approach to survival prediction. Features employed were age, sex, surgical resection status, tumor histology and tumor site, administration of radiation therapy (RT) and chemotherapy status. Concordance index (c-index) was employed to assess the accuracy of survival time prediction. All three models performed well with prediction accuracy (CI 0.767, 0.771, 0.57 for CPH, SVM, RF models respectively) with the best performance achieved when incorporating RT and chemotherapy administration status which emerged as key predictive features. Within the subset of glioblastoma patients, similar prediction accuracy was achieved. These findings should prompt stricter clinician oversight over registry data accuracy through quality assurance as we move towards meaningful predictive ability using ML approaches in glioma.

Modeling on Steering Feedback Torque Based on Data-Driven Method

The steering feedback torque (SFT) is a key part of driving simulator and steer-by-wire system, which provides driver with desired road feel and vehicle motion dynamics. Therefore, accurately modeling of SFT is of great significance for driver to get better steering feel. Since SFT can be affected by many linear or nonlinear factors, it is appropriate to model SFT using data-driven method. In this article, we adopt artificial neural network (ANN) and Gaussian process regression (GPR) to build the SFT model, and analyze the performance. Considering the fact that the contributing factors for SFT may vary under different driving conditions, we employ K-Means to precluster the training dataset to improve the model accuracy. The model training and validation processes are mainly data-driven, and the results show that GPR and ANN can achieve similar prediction accuracy with the mean square error to be around 0.10. Since the GPR model can be trained much faster than ANN model, it is more suitable for real-time application. It is further demonstrated that using preclustered data based on K-Means for model training can significantly improve its accuracy without sacrificing its computational efficiency.

Towards spatially continuous mapping of soil organic carbon in croplands using multitemporal Sentinel-2 remote sensing

Intensified human activities can augment soil organic carbon (SOC) losses from the world’s croplands, making SOC a highly dynamic parameter both in space and time. Sentinel-2 spectral imagery is well placed to capture the spatiotemporal variability of SOC, but its capability has only been demonstrated for agricultural regions mostly located in Europe. Furthermore, most studies so far only used single-date images that resulted in spatially non-continuous SOC maps, hampering their ability to aid multiscale SOC assessments. Here, we aim to achieve spatially continuous mapping of SOC in croplands, by creating multitemporal bare soil composites that maximize cropland coverage in two regions of varying agroecosystems and landscape structure in the Northeast China Chernozem region and the Belgian Loam Belt. Bare soil pixels were extracted via spectral index thresholding that excluded contaminated pixels from external perturbance. Multitemporal soil composites were then obtained by averaging over multiple single-date bare soil images that were selected within pre-determined optimal time-windows, corresponding to the region-specific crop sowing periods when best possible surface conditions were expected. Results show that the optimal time-window filter ensured selective inclusion of single-date images that themselves yielded stable and robust SOC predictions across multiple years. Spectral-based models developed from multitemporal composites consistently produced better or similar prediction accuracies than single-date images for both study regions (R2: 0.52–0.62; RMSE: 0.17–0.21 g 100 g−1), while also achieved maximum cropland coverage (>82 %). Bootstrap modelling demonstrated that SOC mapping via multitemporal Sentinel-2 data was associated with small uncertainties. Investigations into the significant spectral bands that contributed to the prediction of SOC suggested that, regardless of the study regions, the physical relationship between spectral bands and SOC that predominantly exists for laboratory spectra is largely translated into Sentinel-2 platforms. This study highlights the widespread applicability of multitemporal Sentinel-2 remote sensing for effective and high-resolution SOC mapping, in order to detect localized soil degradation as well as to inform regional cropland management in diverse agroecosystems.