Multivariate Adaptive Regression Splines Research Articles

Degrees of Freedom: Search Cost and Self-consistency

Model degrees of freedom ( df ) is a fundamental concept in statistics because it quantifies the flexibility of a fitting procedure and is indispensable in model selection. To investigate the gap between df and the number of independent variables in the fitting procedure, Tibshirani (2015) introduced the search degrees of freedom ( sdf ) concept to account for the search cost during model selection. However, this definition has two limitations: it does not consider fitting procedures in augmented spaces and does not use the same fitting procedure for sdf and df . We propose a modified search degrees of freedom ( msdf ) to directly account for the cost of searching in either original or augmented spaces. We check this definition for various fitting procedures, including classical linear regressions, spline methods, adaptive regressions (the best subset and the lasso), regression trees, and multivariate adaptive regression splines (MARS). In many scenarios when sdf is applicable, msdf reduces to sdf . However, for certain procedures like the lasso, msdf offers a fresh perspective on search costs. For some complex procedures like MARS, the df has been pre-determined during model fitting, but the df of the final fitted procedure might differ from the pre-determined one. To investigate this discrepancy, we introduce the concepts of nominal df and actual df , and define the property of self-consistency, which occurs when there is no gap between these two df ’s. We propose a correction procedure for MARS to align these two df ’s, demonstrating improved fitting performance through extensive simulations and two real data applications.

Improving the predictive accuracy of production frontier models for efficiency measurement using machine learning: The LSB-MAFS method

Making accurate predictions of the true production frontier is critical for reliable efficiency analysis. However, canonical deterministic methods like Data Envelopment Analysis (DEA) provide approximations of the production frontier that cannot accommodate noise satisfactorily and suffer from overfitting. This study combines machine learning techniques known as Least Squares Boosting (LSB) and Multivariate Adaptive Regression Splines (MARS), to introduce a new methodology that improves the accuracy of production frontiers predictions and overcomes previous limitations. The new method fits pairwise regression splines to the data while ensuring that the predicted production frontiers satisfy certain the required regularity conditions: envelopmentness, monotonicity, and concavity. The method, termed LSB-MAFS, is implemented through computational algorithms, and we illustrate its applicability by performing simulations with several data generating processes. We also compare its performance against the most popular alternatives, considering both deterministic and stochastic scenarios: DEA, bootstrapped DEA, Corrected Concave Non-Parametric Least Squares (C2NLS) and Stochastic Frontier Analysis (SFA). The new method outperforms these alternatives in the most complex scenarios, including stochastic settings where parametric methods like SFA should perform better in principle. We conclude that our approach to production frontier prediction is a valid and competitive alternative for dependable efficiency analysis.

Perbandingan Propensity Score Stratification dan Propensity Score Matching dengan Pendekatan Multivariate Adaptive Regression Spline

Research on complications of Diabetes Mellitus (DM) is multifactorial, where the risk factors causing DM complications are interrelated, leading to confounding bias, which results in inaccurate research findings. Confounding bias can be reduced using the propensity score method. This study aims to compare the performance of the Propensity Score Stratification (PSS) and Propensity Score Matching (PSM) methods with the Multivariate Adaptive Regression Spline (MARS) approach in estimating treatment effects on DM complication cases. The data used is the medical records of type-2 DM patients at Hospital X. The results showed that the PSS method with the MARS approach is not suitable for small data sets, as it can lead to treatment or control groups lacking members, making it impossible to calculate the p-value in balance testing or the Percent Bias Reduction (PBR). The estimated Average Treatment Effect (ATE) using the PSS method was 0.487 with a PBR of 35.1%, whereas the estimated Average Treatment for Treated (ATT) using the PSM method was 0.531 with a PBR of 99.46%. These PBR values indicate that the best method for estimating treatment effects and the one that can reduce the most bias in this case is the PSM method with MARS. The analysis also showed that serum uric acid levels significantly affect the peripheral diabetic neuropathy (PDN) status of DM patients.

Black-White Differences in Chronic Stress Exposures to Predict Preterm Birth: Interpretable, Race/Ethnicity-Specific Machine Learning Models.

We developed Multivariate Adaptive Regression Splines (MARS) machine learning models of chronic stressors using the Pregnancy Risk Assessment Monitoring System data (2012-2017) to predict preterm birth (PTB) more accurately and identify chronic stressors driving PTB among non-Hispanic (N-H) Black and N-H White pregnant women in the U.S. We trained the MARS models using 5-fold cross-validation, whose performance was evaluated with AUC. We computed variable importance for PTB prediction. Our models showed high accuracy (AUC: 0.754-0.765). The number of prenatal care visits, premature rupture of membrane, and medical conditions were the most important variables in predicting PTB across the populations. Chronic stressors (e.g., low maternal education and violence) and their correlates were pivotal for PTB prediction only for N-H Black women. Interpretable, race/ethnicity-specific MARS models can predict PTB accurately and explain the most impactful life stressors and their magnitude of effect on PTB risk among N-H Black and N-H White women.

Enhancing genomic prediction with Stacking Ensemble Learning in Arabica Coffee.

Coffee Breeding programs have traditionally relied on observing plant characteristics over years, a slow and costly process. Genomic selection (GS) offers a DNA-based alternative for faster selection of superior cultivars. Stacking Ensemble Learning (SEL) combines multiple models for potentially even more accurate selection. This study explores SEL potential in coffee breeding, aiming to improve prediction accuracy for important traits [yield (YL), total number of the fruits (NF), leaf miner infestation (LM), and cercosporiosis incidence (Cer)] in Coffea Arabica. We analyzed data from 195 individuals genotyped for 21,211 single-nucleotide polymorphism (SNP) markers. To comprehensively assess model performance, we employed a cross-validation (CV) scheme. Genomic Best Linear Unbiased Prediction (GBLUP), multivariate adaptive regression splines (MARS), Quantile Random Forest (QRF), and Random Forest (RF) served as base learners. For the meta-learner within the SEL framework, various options were explored, including Ridge Regression, RF, GBLUP, and Single Average. The SEL method was able to predict the predictive ability (PA) of important traits in Coffea Arabica. SEL presented higher PA compared with those obtained for all base learner methods. The gains in PA in relation to GBLUP were 87.44% (the ratio between the PA obtained from best Stacking model and the GBLUP), 37.83%, 199.82%, and 14.59% for YL, NF, LM and Cer, respectively. Overall, SEL presents a promising approach for GS. By combining predictions from multiple models, SEL can potentially enhance the PA of GS for complex traits.

Prediction of visceral adipose tissue magnitude using a new model based on simple clinical measurements.

Waist circumference (WC) is a reliable obesity surrogate but may not distinguish between visceral and subcutaneous adipose tissue. Our aim was to develop a novel sex-specific model to estimate the magnitude of visceral adipose tissue measured by computed tomography (CT-VAT). The model was initially formulated through the integration of anthropometric measurements, laboratory data, and CT-VAT within a study group (n=185), utilizing the Multivariate Adaptive Regression Splines (MARS) methodology. Subsequently, its correlation with CT-VAT was examined in an external validation group (n=50). The accuracy of the new model in estimating increased CT-VAT (>130 cm2) was compared with WC, body mass index (BMI), waist-hip ratio (WHR), visceral adiposity index (VAI), a body shape index (ABSI), lipid accumulation product (LAP), body roundness index (BRI), and metabolic score for visceral fat (METS-VF) in the study group. Additionally, the new model's accuracy in identifying metabolic syndrome was evaluated in our Metabolic Healthiness Discovery Cohort (n=430). The new model comprised WC, gender, BMI, and hip circumference, providing the highest predictive accuracy in estimating increased CT-VAT in men (AUC of 0.96 ± 0.02), outperforming other indices. In women, the AUC was 0.94 ± 0.03, which was significantly higher than that of VAI, WHR, and ABSI but similar to WC, BMI, LAP, BRI, and METS-VF. It's demonstrated high ability for identifying metabolic syndrome with an AUC of 0.76 ± 0.03 (p<0.001). The new model is a valuable indicator of CT-VAT, especially in men, and it exhibits a strong predictive capability for identifying metabolic syndrome.

On-farm cereal rye biomass estimation using machine learning on images from an unmanned aerial system

This study assesses the potential of using multispectral images collected by an unmanned aerial system (UAS) on machine learning (ML) frameworks to estimate cereal rye (Secale cereal L.) biomass. Multispectral images and ground-truth cereal rye biomass data were collected from 15 farmers’ fields up to three times between March and May in northwest Ohio. Images were processed to derive 13 vegetation indices (VIs). Out of 13 VIs, six optimal sets of VIs, including excess green (ExG), normalized green red difference index (NGRDI), soil adjusted vegetation index (SAVI), blue green ratio (B_G_ratio), red-edge triangular vegetation index (RTVI), and normalized difference red-edge (NDRE) were selected using the variance inflation factor (VIF) based feature selection approach. Six regression models including a multiple linear regression (MLR), elastic net (ENET), multivariate adaptive regression splines (MARS), support vector machine (SVM), random forest (RF), and extreme gradient boosting (XGB) were investigated for estimation of cereal rye biomass based on the VIs. For most of the models, the six selected VIs performed better than or similar to the full set of 13 VIs with R2 ranging from 0.24 to 0.59 and RMSE ranging from 83.13 to 91.89 g/m2 during 10-fold cross-validation. During independent accuracy assessment with the selected set of VIs, XGB exhibited the highest R2 (0.67) and lowest RMSE (83.13 g/m2) and MAE (48.13 g/m2) followed by RF and ENET. For all the models, the agreement between observed and predicted biomass was high for biomass less than or equal to 200 g/m2 but decreased for biomass greater than 200 g/m2. When field-collected structural features were integrated with the selected VIs, the models showed improved performance, with R2 and RMSE of the models reaching up to 0.82 and 61.67 g/m2 respectively. Among the six VIs, SAVI showed the strongest impact on the model prediction for the best-performing RF and XGB regression models. The findings of this study demonstrate the potential of precisely estimating and mapping cereal rye biomass based on UAS-captured multispectral images. Timely information on cover crop growth can facilitate numerous decision-making processes, including planning the planting operations, and management of nutrients, weeds, and soil moisture to improve agronomic and environmental outcomes.

Glioma Grade and Molecular Markers: Comparing Machine-Learning Approaches Using VASARI (Visually AcceSAble Rembrandt Images) Radiological Assessment.

This study aimed to leverage Visually AcceSAble Rembrandt Images (VASARI) radiological features, extracted from magnetic resonance imaging (MRI) scans, and machine-learning techniques to predict glioma grade, isocitrate dehydrogenase (IDH) mutation status, and O6-methylguanine-DNA methyltransferase (MGMT) methylation. A retrospective evaluation was undertaken, analyzing MRI and molecular data from 107 glioma patients treated at a tertiary hospital. Patients underwent MRI scans using established protocols and were evaluated based on VASARI criteria. Tissue samples were assessed for glioma grade and underwent molecular testing for IDH mutations and MGMT methylation. Four machine learning models, namely, Random Forest, Elastic-Net, multivariate adaptive regression spline (MARS), and eXtreme Gradient Boosting (XGBoost), were trained on 27 VASARI features using fivefold internal cross-validation. The models' predictive performances were assessed using the area under the curve (AUC), sensitivity, and specificity. For glioma grade prediction, XGBoost exhibited the highest AUC (0.978), sensitivity (0.879), and specificity (0.964), with f6 (proportion of non-enhancing) and f12 (definition of enhancing margin) as the most important predictors. In predicting IDH mutation status, XGBoost achieved an AUC of 0.806, sensitivity of 0.364, and specificity of 0.880, with f1 (tumor location), f12, and f30 (perpendicular diameter to f29) as primary predictors. For MGMT methylation, XGBoost displayed an AUC of 0.580, sensitivity of 0.372, and specificity of 0.759, highlighting f29 (longest diameter) as the key predictor. This study underscores the robust potential of combining VASARI radiological features with machine learning models in predicting glioma grade, IDH mutation status, and MGMT methylation. The best and most balanced performance was achieved using the XGBoost model. While the prediction of glioma grade showed promising results, the sensitivity in discerning IDH mutations and MGMT methylation still leaves room for improvement. Follow-up studies with larger datasets and more advanced artificial intelligence techniques can further refine our understanding and management of gliomas.

Prediction equations for detecting COVID-19 infection using basic laboratory parameters.

Coronavirus disease 2019 (COVID-19) emerged as a global pandemic during 2019 to 2022. The gold standard method of detecting this disease is reverse transcription-polymerase chain reaction (RT-PCR). However, RT-PCR has a number of shortcomings. Hence, the objective is to propose a cheap and effective method of detecting COVID-19 infection by using machine learning (ML) techniques, which encompasses five basic parameters as an alternative to the costly RT-PCR. Two machine learning-based predictive models, namely, Artificial Neural Network (ANN) and Multivariate Adaptive Regression Splines (MARS), are designed for predicting COVID-19 infection as a cheaper and simpler alternative to RT-PCR utilizing five basic parameters [i.e., age, total leucocyte count, red blood cell count, platelet count, C-reactive protein (CRP)]. Each of these parameters was studied, and correlation is drawn with COVID-19 diagnosis and progression. These laboratory parameters were evaluated in 171 patients who presented with symptoms suspicious of COVID-19 in a hospital at Kharagpur, India, from April to August 2022. Out of a total of 171 patients, 88 and 83 were found to be COVID-19-negative and COVID-19-positive, respectively. The accuracies of the predicted class are found to be 97.06% and 91.18% for ANN and MARS, respectively. CRP is found to be the most significant input parameter. Finally, two predictive mathematical equations for each ML model are provided, which can be quite useful to detect the COVID-19 infection easily. It is expected that the present study will be useful to the medical practitioners for predicting the COVID-19 infection in patients based on only five very basic parameters.

Machine learning risk stratification for high-risk infant follow-up of term and late preterm infants.

Term and late preterm infants are not routinely referred to high-risk infant follow-up programs at neonatal intensive care unit (NICU) discharge. We aimed to identify NICU factors associated with abnormal developmental screening and develop a risk-stratification model using machine learning for high-risk infant follow-up enrollment. We performed a retrospective cohort study identifying abnormal developmental screening prior to 6 years of age in infants born ≥34 weeks gestation admitted to a level IV NICU. Five machine learning models using NICU predictors were developed by classification and regression tree (CART), random forest, gradient boosting TreeNet, multivariate adaptive regression splines (MARS), and regularized logistic regression analysis. Performance metrics included sensitivity, specificity, accuracy, precision, and area under the receiver operating curve (AUC). Within this cohort, 87% (1183/1355) received developmental screening, and 47% had abnormal results. Common NICU predictors across all models were oral (PO) feeding, follow-up appointments, and medications prescribed at NICU discharge. Each model resulted in an AUC > 0.7, specificity >70%, and sensitivity >60%. Stratification of developmental risk in term and late preterm infants is possible utilizing machine learning. Applying machine learning algorithms allows for targeted expansion of high-risk infant follow-up criteria. This study addresses the gap in knowledge of developmental outcomes of infants ≥34 weeks gestation requiring neonatal intensive care. Machine learning methodology can be used to stratify early childhood developmental risk for these term and late preterm infants. Applying the classification and regression tree (CART) algorithm described in the study allows for targeted expansion of high-risk infant follow-up enrollment to include those term and late preterm infants who may benefit most.

Creating a Lactation Model for 305-Day Milk Yield with Different Resampling Techniques (Bagging Mars) in Mars Modeling

The main purpose of this research is to obtain a prediction model for milk yield by using Multivariate Adaptive Regression Splines (MARS) and Bagging MARS algorithms as a non-parametric regression technique. For this purpose, the effects on milk yield of 305 days were investigated by using lactation parameters in dairy cattle. In the study, 9337 lactation milk yield records belonging to 37 animals belonging to the 2022-2023 period were used and the data set was created by randomly ordering the animals. Data on milk yield results were analyzed with MARS and Bagging MARS algorithms. For dairy cattle; it was modeled with explanatory variables such as lactation month (month), service period (SP), last 7 days average milk yield (L7DMMY), animal's first birth age (FP), animal's age (Age), number of lactations (LN).Correlation coefficient (r), coefficient of determination (R2), Adjusted R2, Root of Square Mean Error (RMSE), standard deviation ratio (SD ratio), mean absolute percent error (MAPE), mean absolute for MARS algorithm estimating total average milk yield deviation (MAD) and Akaike Information Criteria (AIC) values are 0.9986, 0.997, 0.977, 0.142, 0.052, 0.2389, 0.086 and -88, respectively. Similar statistics for the Bagging MARS algorithm are 0.754, 0.556, 0.453, 1.8, 0.666, 3.96, 1.47, and 115, respectively. It has been observed that MARS and Bagging MARS algorithms provide correct results according to the goodness of fit statistics. In this study, it was revealed that MARS algorithm gave better results in milk yield modeling of 305-day lactation.

Traditional preference mapping and computational machine learning techniques: A comparative study of approaches to guide product development

Preference mapping, a well-known set of multivariate statistical techniques, has become widely adopted due to its demonstrated effectiveness as a powerful tool in guiding the development of new products and enhancing existing ones. Recent advancements in open-source software and computational capabilities have introduced a new set of accessible tools with the potential to address limitations associated with traditional methods. This study introduces an alternative algorithm for building predictive models, employing regularized regression in combination with Multivariate Adaptive Regression Spline (MARS). These methods make fewer assumptions about the relationship between predictors and the target and can easily capture complex and non-linear relationships. Additionally, the study presents a robust and systematic alternative approach for calculating optimum profiles and performing simulations. The paper aims to compare this new set of tools, referred to as computational machine learning techniques, with a well-established and widely recognized method − PrefMap based on Partial Least Squares Regression. The primary intention of the comparison between computational machine learning and one example of a traditional approach is not to determine a winning methodology, but rather to enhance awareness and deepen the understanding of this emerging family of models and techniques now available to sensory and consumer scientists. Results are assessed side by side to reveal their similarities and differences in terms of predictive power, drivers of liking, and the optimal profile aspects, and a list of practical considerations is provided at the end, enabling a better understanding of the trade-offs between the two approaches presented here.

Slope Stability Analysis of Vetiver Grass Stabilized Soil Using Genetic Programming and Multivariate Adaptive Regression Splines

Slope Stability Analysis of Vetiver Grass Stabilized Soil Using Genetic Programming and Multivariate Adaptive Regression Splines

Analysis of vacuum operation on hydrogen separation from H2/H2O mixture via Pd membrane using Taguchi method, response surface methodology, and multivariate adaptive regression splines

The influence of vacuum pressure applied on H2 separation from a palladium membrane is explored in this study. Three factors with three levels are considered, including the membrane chamber temperature with levels 320 °C, 350 °C, and 380 °C; the retentate-side total pressure with levels 1, 2, and 3 atm; and the permeation-side vacuum pressure with levels 0, 25, and 50 kPa. The Taguchi, response surface methodology (RSM), and multivariate adaptive regression splines (MARS) methods are employed to analyze the effects of the three parameters on hydrogen separation and predict their optimal combination. The retentate-side total pressure exhibits the highest impact on H2 permeation, following the permeation-side vacuum pressure and then the membrane chamber temperature. The maximum H2 flux is 0.226 mol∙s−1∙m−2, with H2 recovery of 91 % obtained at the optimal conditions with a temperature of 380 °C, a total pressure of 3 atm, and a vacuum pressure of 50 kPa. The improvement in H2 flux reaches 21.6 % compared with the case without the imposed vacuum pressure at the same temperature and total pressure. This result shows the imposed vacuum pressure is an efficient way to enhance H2 permeation. The maximum relative errors between the experimental data and the predictions from the Taguchi, RSM, and MARS methods are 6.74 %, 3.37 %, and 8.08 %, respectively. The RSM method presents higher accuracy than Taguchi and MARS, perhaps due to a more precise analysis of the interaction terms. The smaller amount of input data and ignoring the temperature effect in MARS could be the reason for the lower accuracy. Nevertheless, the MARS method still demonstrates acceptable results. The cost of the Taguchi method is lower than that of the RSM method since it requires fewer experimental cases. In a word, the choice of the prediction method depends on the desired accuracy and the experimental cost.

Coupled extreme gradient boosting algorithm with artificial intelligence models for predicting compressive strength of fiber reinforced polymer- confined concrete

Accurately predicting and identifying appropriate parameters are necessary for producing a safe and reliable strength model of concrete elements confined with fiber-reinforced polymers (FRP). In this study, an extreme gradient boosting (XGBoost) algorithm was developed for the feature selection and prediction of the ultimate compressive strength of FRP-confined concrete. The modeling process was established using a dataset from open-source literature consisting of 490 circular columns. Three well-known artificial intelligence (AI) models, the multivariate adaptive regression spline (MARS), extreme learning machine (ELM), and RANdom Forest GEnRator (Ranger), were used to validate the proposed model. The results demonstrated the effectiveness of the XGBoost algorithm in the modeling process, selection of suitable parameters, and enhancement of the prediction accuracy. The algorithm achieved excellent prediction results for all input combinations with a coefficient of determination (R2) greater than 0.9, and the best performance is gained by using five input parameters with (R2 = 0.955), mean absolute percentage error (MAPE = 0.130), and root mean square error (RMSE = 0.572). The study revealed the flexibility and efficiency of capturing the nonlinear behavior of complex FRP-confined concrete using the proposed model.

Employing machine learning (ML) and explainable artificial intelligence (XAI) to predict and explain suicidal ideation among patients with prostate cancer.

e23086 Background: It is anticipated that implementing machine learning (ML) strategies will impact the clinical decision-making process in the future. This study aimed to compare the accuracy of various machine learning algorithms in predicting suicidal ideation in patients who were receiving treatment for prostate cancer (PC). Further, to predict suicidal ideation and explain the burden of illness (BOI) by utilizing ML algorithms and XAI-based interpretability. Methods: We analyzed the United States 2017 National Inpatient Sample database for patients hospitalized for PC using linear and non-linear ML methods. Using techniques such as forward selection (FS) and backward elimination (BE), Random Forest (RF), decision trees, Multivariate Adaptive Regression Splines, and Gradient Boosting Machine (GBM), we determined subsets and features. We used linear and non-linear MLs-- Lasso, Ridge, RF, and Neural Networks (NN). A 70% and 30% partitioning was performed on the training and test datasets. The performance of the model was tested using discrimination (C-statistics), the Receiver-Operating Characteristics (ROC) curve, and the Hosmer-Lemeshow tests on both the training data and the test data. Several XAI approaches, including permutation significance, global surrogate marker, feature interpretation, interactivity, and local interpretability, were also utilized to examine the BOI (Length of stay (LOS)) among PC cancer cohort with suicide ideation. Results: We identified 680 patients with suicidal ideation among 208,730 PC patients. The most important variable derived through FS and BE were depression, drug abuse, psychiatric disorder, alcohol dependence, race, Medicaid beneficiaries, cardiac arrhythmias, metastasis, weight loss, congestive heart failure, and un-complicated hypertension, further confirmed through the GBM and other methods. There was successful documentation of demographic, socioeconomic, and clinical characteristics through feature selection approaches. Linear models, Lasso, and Ridge demonstrated an excellent area under the ROC for all cohorts ( &gt; 0.9 in train and test datasets). On the other hand, tree-based models, such as RF, performed poorly (AUC 0.72 for train and 0.66 for test). The findings of the XAI methodologies showed that among PC suicide ideation, higher BOI was associated with patients staying in low-income neighborhoods, Whites, Medicaid beneficiaries, and those with weight loss. Conclusions: We showcase the increasing capability of predictive analytics, particularly XAI, in predicting suicidal ideation. The combination of the vast amount of data generated by the digitalization of healthcare systems with the advanced processing power of modern servers enables the development of machine learning-based care pathways, which has the potential to revolutionize clinical decision-making in cancer.

In-depth simulation of rainfall–runoff relationships using machine learning methods

ABSTRACT Measurement inaccuracies and the absence of precise parameters value in conceptual and analytical models pose challenges in simulating the rainfall–runoff modeling (RRM). Accurate prediction of water resources, especially in water scarcity conditions, plays a distinctive and pivotal role in decision-making within water resource management. The significance of machine learning models (MLMs) has become pronounced in addressing these issues. In this context, the forthcoming research endeavors to model the RRM utilizing four MLMs: Support Vector Machine, Gene Expression Programming (GEP), Multilayer Perceptron, and Multivariate Adaptive Regression Splines (MARS). The simulation was conducted within the Malwathu Oya watershed, employing a dataset comprising 4,765 daily observations spanning from July 18, 2005, to September 30, 2018, gathered from rainfall stations, and Kappachichiya hydrometric station. Of all input combinations, the model incorporating the input parameters Qt−1, Qt−2, and R̄t was identified as the optimal configuration among the considered alternatives. The models' performance was assessed through root mean square error (RMSE), mean average error (MAE), coefficient of determination (R2), and developed discrepancy ratio (DDR). The GEP model emerged as the superior choice, with corresponding index values (RMSE, MAE, R2, DDRmax) of (43.028, 9.991, 0.909, 0.736) during the training process and (40.561, 10.565, 0.832, 1.038) during the testing process.

Current and future scenarios of suitability and expansion of cassava brown streak disease, Bemisia tabaci species complex, and cassava planting in Africa.

Cassava (Manihot esculenta) is among the most important staple crops globally, with an imperative role in supporting the Sustainable Development Goal of 'Zero hunger'. In sub-Saharan Africa, it is cultivated mainly by millions of subsistence farmers who depend directly on it for their socio-economic welfare. However, its yield in some regions has been threatened by several diseases, especially the cassava brown streak disease (CBSD). Changes in climatic conditions enhance the risk of the disease spreading to other planting regions. Here, we characterise the current and future distribution of cassava, CBSD and whitefly Bemisia tabaci species complex in Africa, using an ensemble of four species distribution models (SDMs): boosted regression trees, maximum entropy, generalised additive model, and multivariate adaptive regression splines, together with 28 environmental covariates. We collected 1,422 and 1,169 occurrence records for cassava and Bemisia tabaci species complex from the Global Biodiversity Information Facility and 750 CBSD occurrence records from published literature and systematic surveys in East Africa. Our results identified isothermality as having the highest contribution to the current distribution of cassava, while elevation was the top predictor of the current distribution of Bemisia tabaci species complex. Cassava harvested area and precipitation of the driest month contributed the most to explain the current distribution of CBSD outbreaks. The geographic distributions of these target species are also expected to shift under climate projection scenarios for two mid-century periods (2041-2060 and 2061-2080). Our results indicate that major cassava producers, like Cameron, Ivory Coast, Ghana, and Nigeria, are at greater risk of invasion of CBSD. These results highlight the need for firmer agricultural management and climate-change mitigation actions in Africa to combat new outbreaks and to contain the spread of CBSD.

Prediction of the Global Distribution of Arhopalus rusticus under Future Climate Change Scenarios of the CMIP6

Arhopalus rusticus is a significant forestry pest known for its destructive impact on various host plants. This species, commonly found in coniferous forests across the Northern Hemisphere, has successfully spread to regions like New Zealand, Australia, and South America. This research is based on the known distribution sites of A. rusticus. Projections are made for the potential global distribution of A. rusticus under historical climatic conditions (1970–2000) and future climatic conditions (2081–2100) for the four forcing scenarios of the Coupled Model International Comparison Program 6 (CMIP6). The aim was to analyze the effects of climate change on the distribution range of this pest and its invasion trend in the southern hemisphere, and to support relevant departments in enhancing the effectiveness of forestry pest control strategies. The study utilized the Biomod2 software package in R to compare six models: generalized linear models (GLMs), generalized additive models (GAMs), multivariate adaptive regression splines (MARSs), artificial neural networks (ANNs), classification and regression trees (CTAs), and random forests (RFs) for modeling species distributions. The optimal model was selected based on evaluation indexes such as AUC and TSS. Projections of A. rusticus distribution under historical and future climate scenarios were created. The prediction results were visualized using ArcGIS software (version 10.2) to classify fitness levels and calculate distribution areas. Based on evaluation metrics, random forests (RFs) demonstrated the highest average assessment index scores, indicating high prediction accuracy (AUC = 0.99, TSS = 0.91, Kappa = 0.93). Model predictions revealed that, under historical climatic conditions, A. rusticus was predominantly found in northern Europe, eastern Asia, eastern and southwestern coastal regions of North America, and there were also highly suitable regions in parts of the southern hemisphere, including central and southwestern Argentina, southern Australia, New Zealand, and South Africa. Among these models, each of the CMIP6’s different climate prediction scenarios had a significant impact on the predicted distribution of A. rusticus. The SSP126 scenario depicted the broadest range of suitability, while the SSP585 scenario presented the narrowest and, overall, the extent of highly suitable regions was contracting. Multi-model predictions suggested that the potential distribution area of A. rusticus during the period of 2081–2100 would likely expand compared to that of 1970–2000, ranging from an increase of 1.13% (SSP126) up to 6.61% (SSP585), positively correlating with the level of radiative forcing. Notably, the most substantial growth was observed in potentially low-suitability region, escalating from 1.17% (SSP126) to 5.55% (SSP585). The distribution of A. rusticus shows decreasing trends from coastal areas to inland areas and from high to low level suitability of regions, and further expansion into the southern hemisphere under future climate conditions. Therefore, quarantine efforts at ports of entry should be strengthened in areas that are not currently infested but are at risk of invasion, and precise preventive measures should be strengthened in areas that are at risk of further expansion under future climatic conditions to prevent its spread to inland areas.

Unveiling the nexus between atmospheric visibility, remotely sensed pollutants, and climatic variables across diverse topographies: A data-driven exploration empowered by artificial intelligence

Deteriorating visual range (VR) can cause challenges for the transportation sector, resulting in economic and life losses. Air pollutants, smoke, fog, and many meteorological parameters such as air temperature (T), relative humidity (RH), wind speed (WS), and wind direction (WD) can contribute to light extinction and degrade VR. Advancements in geospatial technologies have triggered artificial intelligence to analyze and model the relationships among environmental and climatological parameters. This paper aims to assess the potential of supervised machine learning models for the parameterization of VR over Pakistan's diverse topography by utilizing meteorological parameters and some pollutants. The daily data from 2005 to 2020 of VR, T, RH, WS, WD, Aerosol Optical Depth (AOD), Nitrogen dioxide (NOx), Sulfate, Sulfur dioxide (SOx), and Dust were acquired. Ten machine learning models, including Random Forest (RF), Extreme Gradient Boosting (XGB), Artificial Neural Networks (ANN), Support Vector Machine (SVM), Decision Trees (DT), Gradient Boosting Machine (GBM), Causal, Unbiased, Binned, and Intermittent, Search, and Tree (CUBIST), Multi-Layer Perceptron (MLP), Multivariate Adaptive Regression Splines (MARS), and K-Nearest Neighbor (KNN) were gauged for VR estimation. We also coupled the Bagged Extreme Gradient Boosting (BG-XG) model by combining XGB and bagging technique. BG-XG performed better than the rest of the models, with coefficients of determination of 0.90 for the training and 0.70 to 0.90 for the validation set. Degradation in the VR was highly dependent on the changes in RH followed by SOx and dust associated with anthropogenic emissions. RH, SO4, and SO2 emerged as the most important parameters for the VR decline. Proposed model parameters can be helpful in accurate VR projections and improving severe weather alerts, including analyzing and managing air pollution. This work will also be helpful to improve aviation and transportation safety for pilots, drivers, and automated vehicles to minimize low-visibility accidents.