Mean Decrease Accuracy Research Articles

Perfluorooctanoic Acids (PFOA) removal using electrochemical oxidation: A machine learning approach

The urgent need to eliminate Perfluorooctanoic Acid (PFOA) has positioned electrooxidation (EO) as a key solution for pollutant degradation. This study evaluates several machine learning (ML) models, including K-Nearest Neighbors (KNN), Decision Tree (DT), Random Forest (RF), Gradient Boosted Decision Trees (GBDT), and Deep Learning (DL), to predict EO efficiency in PFOA removal. Using 10-fold cross-validation, the RF model outperformed others with a root mean square error (RMSE) of 7.7 and a correlation coefficient of 0.965, demonstrating its robustness and accuracy across diverse operational settings. Feature importance within the RF model was analyzed using Gini impurity and Mean Decrease in Accuracy (MDA). Electrolysis Time consistently emerged as the most influential factor in both analyses, underscoring its pivotal role in providing extended exposure of PFOA molecules to reactive species at the electrode surfaces. The study also found strong agreement between Gini and MDA in identifying Current Density and Anode Material as critical factors, although MDA placed slightly more emphasis on Anode Material. Differences between Gini and MDA were more pronounced in the ranking of Electrolyte Type and Concentration, with MDA assigning higher importance to Electrolyte Concentration. In contrast, the Water Matrix was consistently ranked as the least important factor. The strong concordance between Gini and MDA highlights the reliability of the RF model in identifying key drivers of electrochemical degradation. Overall, this work contributes significantly to the advancement of pollutant degradation technologies, presenting a reliable ML-based tool for environmental remediation efforts.

Mapping dominant tree species of miombo woodlands in Western Tanzania using PlanetScope imagery

Mapping dominant tree species in miombo woodlands is essential for enhancing their monitoring and management. We evaluated PlanetScope imagery to map Julbernardia globiflora, Brachystegia spiciformis, and Pterocarpus tinctorius in Tongwe Forest Reserve, Tanzania. The study assessed the effectiveness of PlanetScope bands in discriminating tree species and investigated how different months/seasons influenced tree species classification. Optimal months (seasons) and spectral bands were selected using Principal Component loading, temporal pattern analysis, mean decrease in accuracy, and mean decrease Gini techniques. Random forest classification was employed for tree species classification, and accuracy was assessed using an error matrix. The study identified March, July, and September as optimal months for acquiring imagery, with effective bands including blue, green-1, green, yellow, red, and red-edge. July and September imagery in the dry season achieved higher overall accuracies of 65% and 69%, respectively, while March imagery in the wet season reached 55%. The highest overall accuracy of 72% was achieved using images from different seasons. Producer’s accuracy was highest for Brachystegia spiciformis (79%) and Julbernardia globiflora (95%), whereas Pterocarpus tinctorius had lower accuracy (25%). User’s accuracy varied with 74% for Brachystegia spiciformis, 70% for Julbernardia globiflora, and 67% for Pterocarpus tinctorius. Mapping accuracy was notably high for Brachystegia spiciformis and Julbernardia globiflora, reflecting their high sample size (dominance) and distinct phenology. The yellow and red bands were particularly effective in distinguishing miombo tree species demonstrating PlanetScope’s capability. Future research should focus on scaling up PlanetScope’s application for broad miombo tree species mapping.

Identification of blood lipid markers of docetaxel treatment in prostate cancer patients

Docetaxel is commonly used for treatment of castration-resistant prostate cancer. Unfortunately, many prostate cancer patients develop resistance to docetaxel. Clinical markers less invasive than biopsies, such as blood samples, would be ideal for monitoring and predicting patient treatment outcomes to docetaxel. Lipid alterations are often associated with the progression of many cancers, including prostate cancer. This study investigated the use of lipids from whole blood as clinical markers for docetaxel resistance in a small cohort of patients with prostate cancer. Qualitative lipidomics was performed by liquid chromatography-tandem mass spectrometry to assess the lipid composition of prostate cancer cells exposed to docetaxel as well as whole blood from prostate cancer patients before, during and after docetaxel treatment. Three patients had castration resistant prostate cancer, three had castration sensitive prostate cancer, and four had de novo prostate cancer during the extent of the study. Mean decrease accuracy and classical univariate receiving operating characteristic curve analyses were performed to identify potential biomarkers. In total, 245 and 221 altered lipids were identified from a second stage of mass spectrometry analysis of prostate cancer cells and clinical blood samples, respectively. Both models indicated that docetaxel treatment altered ether-linked phosphatidylcholines, lysophosphatidylcholine, diacylglycerols, ceramides, hexosylceramides, and sphingomyelins. The results also indicated several lipid changes were associated with sphingolipid signaling and metabolism, and glycerophospholipid metabolism. Collectively, these data suggest the potential usage of identified lipid species as indicators of docetaxel resistance in prostate cancer.

Prognosis Prediction in Head and Neck Squamous Cell Carcinoma by Radiomics and Clinical Information.

(1) Background: head and neck squamous cell carcinoma (HNSCC) is a common cancer whose prognosis is affected by its heterogeneous nature. We aim to predict 5-year overall survival in HNSCC radiotherapy (RT) patients by integrating radiomic and clinical information in machine-learning models; (2) Methods: HNSCC radiotherapy planning computed tomography (CT) images with RT structures were obtained from The Cancer Imaging Archive. Radiomic features and clinical data were independently analyzed by five machine-learning algorithms. The results were enhanced through a voted ensembled approach. Subsequently, a probability-weighted enhanced model (PWEM) was generated by incorporating both models; (3) Results: a total of 299 cases were included in the analysis. By receiver operating characteristic (ROC) curve analysis, PWEM achieved an area under the curve (AUC) of 0.86, which outperformed both radiomic and clinical factor models. Mean decrease accuracy, mean decrease Gini, and a chi-square test identified T stage, age, and disease site as the most important clinical factors in prognosis prediction; (4) Conclusions: our radiomic-clinical combined model revealed superior performance when compared to radiomic and clinical factor models alone. Further prospective research with a larger sample size is warranted to implement the model for clinical use.

Association Between Myopia and Pupil Diameter in Preschoolers: Evidence from a Machine Learning Approach Based on a Real-World Large-Scale Dataset.

Previous studies have explored the connections between various ocular biological parameters with myopia. Our previous study also found that pupil data can predict the myopic progression during the interventions for myopia. However, studies exploring the association between pupil diameter and myopia in preschoolers with myopia were lacking. Hence this study was aimed to investigate the association between pupil diameter and myopia in preschoolers with myopia based on a real-world, large-scale dataset. Data containing 650,671 preschoolers were collected from a total of 1943 kindergartens in Shenzhen, China. Refraction and pupil parameters were collected. After data filtering, the occurrence of myopia and its association with age, gender, pupil diameter, and other variables, were analyzed. Random forest (RF) and eXtreme gradient boosting (XGBoost) were selected from seven machine learning algorithms to build the model. The mean decrease accuracy (MDA), mean decrease Gini (MDG), and gain feature importance (GFI) techniques were employed to quantify the importance of pupil diameter and other features. After the assessments, 51,325 valid records with complete pupil data were included, and 3468 (6.76%) were identified as myopia based on the calculated cycloplegic refraction. Preschoolers with myopia presented reduced pupil diameter and greater variation (5.00 ± 0.99mm) compared to non-myopic preschoolers (6.22 ± 0.67mm). A nonlinear relationship was found according to the scatterplots between pupil diameter and refraction (R2 = 0.14). Especially preschoolers with myopia had reduced pupil diameter compared to emmetropic preschoolers, but hyperope did not experience additional pupil enlargement. After adjusting for other covariates, this relationship is still consistent (P < 0.001). XGBoost and RF algorithms presented the highest performance and validated the importance of pupil diameter in myopia. Based on a real-world large-scale dataset, the current study illuminated that preschoolers with myopia had a reduced pupil diameter compared to emmetropic preschoolers with a nonlinear pattern. Machine learning algorithms visualized and validated the pivotal role of pupil diameter in myopia. chictr.org Identifier: ChiCTR2200057391.

Exploring the variation in associations between socioeconomic indicators and non-communicable diseases in the Tromsø Study: an algorithmic approach

Aims: We contribute to the methodological literature on the assessment of health inequalities by applying an algorithmic approach to evaluate the capabilities of socioeconomic variables in predicting the prevalence of non-communicable diseases in a Norwegian health survey. Methods: We use data from the seventh survey of the population based Tromsø Study (2015–2016), including 11,074 women and 10,009 men aged 40 years and above. We apply the random forest algorithm to predict four non-communicable disease outcomes (heart attack, cancer, diabetes and stroke) based on information on a number of social root causes and health behaviours. We evaluate our results using the classification error, the mean decrease in accuracy, partial dependence statistics. Results: Results suggest that education, household income and occupation to a variable extent contribute to predicting non-communicable disease outcomes. Prediction misclassification ranges between 25.1% and 35.4% depending on the non-communicable diseases under study. Partial dependences reveal mostly expected health gradients, with some examples of complex functional relationships. Out-of-sample model validation shows that predictions translate to new data input. Conclusions: Algorithmic modelling can provide additional empirical detail and metrics for evaluating heterogeneous inequalities in morbidity. The extent to which education, income and occupation contribute to predicting binary non-communicable disease outcomes depends on both non-communicable diseases and socioeconomic indicator. Partial dependences reveal that social gradients in non-communicable disease outcomes vary in shape between combinations of non-communicable disease outcome and socioeconomic status indicator. Misclassification rates highlight the extent of variation within socioeconomic groups, suggesting that future studies may improve predictive accuracy by exploring further subpopulation heterogeneity.

Application of confocal laser microscopy for identification of modern and fossil pollen grains, an example in palm Mauritiinae

Confocal scanning laser microscopy (CSLM) is becoming a powerful tool for palynological studies. CSLM allows palynomorph image sectioning, internal and surface structures visualization, and 3D reconstruction at a higher resolution than standard light microscopy without extra processing. CSLM images are suitable for several image analysis techniques that could help improve the accuracy and reproducibility of taxa identification. Here, using the palm subtribe Mauritiinae (Arecaceae: Calamoideae: Lepidocaryeae) as a model group, we identify modern and fossil pollen grains using CSLM images coupled with ImageJ/Fiji 1.54f plugins and machine learning statistical analyses. Modern taxa pollen grains including Lepidocaryum tenue Mart., Mauritia flexuosa L.f., Mauritiella armata (Mart.) Burret and Mauritiella aculeata (Kunth) Burret were obtained from Smithsonian Tropical Research Institute (STRI) pollen collection or herbarium exsiccates. Fossil pollen of Grimsdalea magnaclavata Germeraad et al. 1968, and Mauritiidites franciscoi (van der Hammen) van der Hammen & Garcia de Mutis 1966, both from Miocene, and Mauritia pollen type from Holocene were obtained from STRI collection. We measured nine shape and exine quantitative parameters, and one qualitative parameter (pollen aperture). Pollen volume was the most important variable (28.270 mean decrease accuracy), followed by pollen aperture (15.003), Skewness (13.466), and spine density (10.246). The machine learning analysis, which included CART and Random Forests, correctly identified both fossil and extant grains. CSLM and the quantitative analysis of morphological traits are a new frontier in palynological studies.

Effects of Micro-Topography and Vegetation on Soil Moisture on Fixed Sand Dunes in Tengger Desert, China.

Soil moisture is a key factor in arid ecosystems, with local variations influenced by topography and vegetation. Understanding this relationship is crucial for combating desertification. Employing ANOVA, Mean Decrease Accuracy (MDA) analysis from random forest modeling and Structural Equation Modeling (SEM), this study investigates the distribution of soil moisture and its associations with topographic and vegetative factors across four micro-geomorphic units in the Tengger Desert, China. Significant heterogeneity in soil moisture across various layers and locations, including windward and leeward slopes and the tops and bottoms of dunes, was observed. Soil moisture generally increases from the surface down to 300 cm, with diminishing fluctuations at greater depths. Soil moisture peaks in the surface and middle layers on windward slopes and in deep layers at the bottom of dunes, exhibiting an initial rise and then a decline on windward slopes. Topographic (including slope direction and elevation difference) and vegetation (including shrub and herb coverage) factors significantly influence soil moisture across three depth layers. Topographic factors negatively affect soil moisture directly, whereas vegetation positively influences it indirectly, with shrub and herb abundance enhancing moisture levels. These insights inform ecological management and the formulation of soil moisture-conservation strategies in arid deserts. The study underscores customizing sand-binding vegetation to various micro-geomorphic dune units.

Lipoproteins and metabolites in diagnosing and predicting Alzheimer’s disease using machine learning

BackgroundAlzheimer’s disease (AD) is a chronic neurodegenerative disorder that poses a substantial economic burden. The Random forest algorithm is effective in predicting AD; however, the key factors influencing AD onset remain unclear. This study aimed to analyze the key lipoprotein and metabolite factors influencing AD onset using machine-learning methods. It provides new insights for researchers and medical personnel to understand AD and provides a reference for the early diagnosis, treatment, and early prevention of AD.MethodsA total of 603 participants, including controls and patients with AD with complete lipoprotein and metabolite data from the Alzheimer’s disease Neuroimaging Initiative (ADNI) database between 2005 and 2016, were enrolled. Random forest, Lasso regression, and CatBoost algorithms were employed to rank and filter 213 lipoprotein and metabolite variables. Variables with consistently high importance rankings from any two methods were incorporated into the models. Finally, the variables selected from the three methods, with the participants’ age, sex, and marital status, were used to construct a random forest predictive model.ResultsFourteen lipoprotein and metabolite variables were screened using the three methods, and 17 variables were included in the AD prediction model based on age, sex, and marital status of the participants. The optimal random forest modeling was constructed with “mtry” set to 3 and “ntree” set to 300. The model exhibited an accuracy of 71.01%, a sensitivity of 79.59%, a specificity of 65.28%, and an AUC (95%CI) of 0.724 (0.645–0.804). When Mean Decrease Accuracy and Gini were used to rank the proteins, age, phospholipids to total lipids ratio in intermediate-density lipoproteins (IDL_PL_PCT), and creatinine were among the top five variables.ConclusionsAge, IDL_PL_PCT, and creatinine levels play crucial roles in AD onset. Regular monitoring of lipoproteins and their metabolites in older individuals is significant for early AD diagnosis and prevention.

Machine learning identifies risk factors associated with long-term opioid use in fibromyalgia patients newly initiated on an opioid

ObjectivesFibromyalgia is frequently treated with opioids due to limited therapeutic options. Long-term opioid use is associated with several adverse outcomes. Identifying factors associated with long-term opioid use is the first...

랜덤포레스트를 활용한 초⋅중학생 독서실행 영향 요인 탐색

Objectives The purposes of this study factors affecting elementary and middle school students’ reading practice were explored using the random forest technique. Methods For this purpose, data from the 13th year (sixth grade of elementary school) and the 14th year (first year of middle school) of the Korean Children's Panel were used. Using reading practice as a response variable and 111 explanatory variables from household, parent, child, and school/private education surveys, factors affecting their reading practice were explored based on the importance index, and a partial dependence plot (PDP) was used. We used this to understand the relationship of influence. Results As a result of examining the top 10 Mean Decrease Accuracy (MDA) predicting reading performance, in the case of 6th grade elementary school students, the results were achievement pressure, learning activities, teacher relationship, strict rules and directive atmosphere, overall happiness, academic stress, peer relationships, and school rules. , physical self-image, and self-esteem appeared in that order. In the case of first-year middle school students, the order was learning activities, achievement pressure, school rules, peer relationships, overall happiness, academic stress, strict rules and directive atmosphere, teacher relationship, physical self-image, and teacher support. Conclusions Factors that affect elementary and middle school students' reading practice are generally more influenced by the school environment than their parents, and the fact that parents' achievement pressure, school adaptation, and strict rules and directive atmosphere are predicted to affect reading practice. It can be seen that coercion has an effect to some extent. As for other variables, there was a difference in what was predicted by ‘self-esteem’ in the 6th grade of elementary school and ‘teacher’s support’ in the 1st grade of middle school.

Combining Multitemporal Optical and Radar Satellite Data for Mapping the Tatra Mountains Non-Forest Plant Communities

Climate change is significantly affecting mountain plant communities, causing dynamic alterations in species composition as well as spatial distribution. This raises the need for constant monitoring. The Tatra Mountains are the highest range of the Carpathians which are considered biodiversity hotspots in Central Europe. For this purpose, microwave Sentinel-1 and optical multi-temporal Sentinel-2 data, topographic derivatives, and iterative machine learning methods incorporating classifiers random forest (RF), support vector machines (SVMs), and XGBoost (XGB) were used for the identification of thirteen non-forest plant communities (various types of alpine grasslands, shrublands, herbaceous heaths, mountain hay meadows, rocks, and scree communities). Different scenarios were tested to identify the most important variables, retrieval periods, and spectral bands. The overall accuracy results for the individual algorithms reached RF (0.83–0.96), SVM (0.87–0.93), and lower results for XGBoost (0.69–0.82). The best combination, which included a fusion of Sentinel-1, Sentinel-2, and topographic data, achieved F1-scores for classes in the range of 0.73–0.97 (RF) and 0.66–0.95 (SVM). The inclusion of topographic variables resulted in an improvement in F1-scores for Sentinel-2 data by one–four percent points and Sentinel-1 data by 1%–9%. For spectral bands, the Sentinel-2 10 m resolution bands B4, B3, and B2 showed the highest mean decrease accuracy. The final result is the first comprehensive map of non-forest vegetation for the Tatra Mountains area.

Convolutional Neural Network-Driven Impedance Flow Cytometry for Accurate Bacterial Differentiation.

Impedance flow cytometry (IFC) has been demonstrated to be an efficient tool for label-free bacterial investigation to obtain the electrical properties in real time. However, the accurate differentiation of different species of bacteria by IFC technology remains a challenge owing to the insignificant differences in data. Here, we developed a convolutional neural networks (ConvNet) deep learning approach to enhance the accuracy and efficiency of the IFC toward distinguishing various species of bacteria. First, more than 1 million sets of impedance data (comprising 42 characteristic features for each set) of various groups of bacteria were trained by the ConvNet model. To improve the efficiency for data analysis, the Spearman correlation coefficient and the mean decrease accuracy of the random forest algorithm were introduced to eliminate feature interaction and extract the opacity of impedance related to the bacterial wall and membrane structure as the predominant features in bacterial differentiation. Moreover, the 25 optimized features were selected with differentiation accuracies of >96% for three groups of bacteria (bacilli, cocci, and vibrio) and >95% for two species of bacilli (Escherichia coli and Salmonella enteritidis), compared to machine learning algorithms (complex tree, linear discriminant, and K-nearest neighbor algorithms) with a maximum accuracy of 76.4%. Furthermore, bacterial differentiation was achieved on spiked samples of different species with different mixing ratios. The proposed ConvNet deep learning-assisted data analysis method of IFC exhibits advantages in analyzing a huge number of data sets with capacity for extracting predominant features within multicomponent information and will bring about progress and advances in the fields of both biosensing and data analysis.

Comparative study for coastal aquifer vulnerability assessment using deep learning and metaheuristic algorithms.

Coastal aquifer vulnerability assessment (CAVA) studies are essential for mitigating the effects of seawater intrusion (SWI) worldwide. In this research, the vulnerability of the coastal aquifer in the Lahijan region of northwest Iran was investigated. A vulnerability map (VM) was created applying hydrogeological parameters derived from the original GALDIT model (OGM). The significance of OGM parameters was assessed using the mean decrease accuracy (MDA)method, with the current state of SWI emerging as the most crucial factor for evaluating vulnerability. To optimize GALDIT weights, we introduced the biogeography-based optimization (BBO) and gray wolf optimization (GWO) techniques to obtain to hybrid OGM-BBO and OGM-GWO models, respectively. Despite considerable research focused on enhancing CAVA models, efforts to modify the weights and rates of OGM parameters by incorporating deep learning algorithms remain scarce. Hence, a convolutional neural network (CNN) algorithm was applied to produce the VM. The area under the receiver-operating characteristic curves for OGM-BBO, OGM-GWO, and VMCNN were 0.794, 0.835, and 0.982, respectively. According to the CNN-based VM, 41% of the aquifer displayed very high and high vulnerability to SWI, concentrated primarily along the coastline. Additionally, 32% of the aquifer exhibited very low and low vulnerability to SWI, predominantly in the southern and southwestern regions. The proposed model can be extended to evaluate the vulnerability of various coastal aquifers to SWI, thereby assisting land use planers and policymakers in identifying at-risk areas. Moreover, deep-learning-based approaches can help clarify the associations between aquifer vulnerability and contamination resulting from SWI.

Vegetation and Dormancy States Identification in Coniferous Plants Based on Hyperspectral Imaging Data

Conifers are a common type of plant used in ornamental horticulture. The prompt diagnosis of the phenological state of coniferous plants using remote sensing is crucial for forecasting the consequences of extreme weather events. This is the first study to identify the “Vegetation” and “Dormancy” states in coniferous plants by analyzing their annual time series of spectral characteristics. The study analyzed Platycladus orientalis, Thuja occidentalis and T. plicata using time series values of 81 vegetation indices and 125 spectral bands. Linear discriminant analysis (LDA) was used to identify “Vegetation” and “Dormancy” states. The model contained three to four independent variables and achieved a high level of correctness (92.3 to 96.1%) and test accuracy (92.1 to 96.0%). The LDA model assigns the highest weight to vegetation indices that are sensitive to photosynthetic pigments, such as the photochemical reflectance index (PRI), normalized PRI (PRI_norm), the ratio of PRI to coloration index 2 (PRI/CI2), and derivative index 2 (D2). The random forest method also diagnoses the “Vegetation” and “Dormancy” states with high accuracy (97.3%). The vegetation indices chlorophyll/carotenoid index (CCI), PRI, PRI_norm and PRI/CI2 contribute the most to the mean decrease accuracy and mean decrease Gini. Diagnosing the phenological state of conifers throughout the annual cycle will allow for the effective planning of management measures in conifer plantations.

Multiscale Entropy-Based Feature Extraction for the Detection of Instability Inception in Axial Compressors.

The detection of instability inception is favorable to avoid compressor instability. In this paper, a multiscale entropy-based feature extraction is developed for the detection of the instability inception in axial compressors. Nonlinear and statistical features of the short-time instability inception are extracted by generally combining multiscale entropy and statistical features. First, nonlinear features are extracted by refined composite multiscale entropy to avoid the inaccurate estimation or undefined entropy of multiscale entropy for short time series. Second, the time-domain-based statistical features are chosen to capture more information on instability inception, and the dominant statistical features are determined by random forests implemented with the mean decrease accuracy algorithm at each time scale. The obtained refined composite dominant statistical features are regarded as weighting factors and integrated with the refined composite multiscale entropy to generate a combined feature. Finally, numerical simulation results on two synthetic noise datasets and a compressor instability model dataset are presented to demonstrate the effectiveness, efficiency, and robustness of the combined features under different conditions.

Oral microbiota disorder in GC patients revealed by 2b-RAD-M

BackgroundMicrobiota alterations are linked with gastric cancer (GC). However, the relationship between the oral microbiota (especially oral fungi) and GC is not known. In this study, we aimed to apply 2b-RAD sequencing for Microbiome (2b-RAD-M) to characterize the oral microbiota in patients with GC.MethodsWe performed 2b-RAD-M analysis on the saliva and tongue coating of GC patients and healthy controls. We carried out diversity, relative abundance, and composition analyses of saliva and tongue coating bacteria and fungi in the two groups. In addition, indicator analysis, the Gini index, and the mean decrease accuracy were used to identify oral fungal indicators of GC.ResultsIn this study, fungal imbalance in the saliva and tongue coating was observed in the GC group. At the species level, enriched Malassezia globosa (M. globosa) and decreased Saccharomyces cerevisiae (S. cerevisiae) were observed in saliva and tongue coating samples of the GC group. Random forest analysis indicated that M. globosa in saliva and tongue coating samples could serve as biomarkers to diagnose GC. The Gini index and mean decreases in accuracy for M. globosa in saliva and tongue coating samples were the largest. In addition, M. globosa in saliva and tongue coating samples classified GC from the control with areas under the receiver operating curve (AUCs) of 0.976 and 0.846, respectively. Further ecological analysis revealed correlations between oral bacteria and fungi.ConclusionFor the first time, our data suggested that changes in oral fungi between GC patients and controls may help deepen our understanding of the complex spectrum of the different microbiotas involved in GC development. Although the cohort size was small, this study is the first to use 2b-RAD-M to reveal that oral M. globosa can be a fungal biomarker for detecting GC.

Does rice paddy management increase soil organic carbon in the warm temperate and tropical regions?

Enhancing soil organic carbon (SOC) storage in agricultural fields is an effective option for climate change mitigation and food security. In temperate regions, rice farming under paddy is known to increase SOC stock due mainly to the reduced decomposition caused by prolonged submerged conditions compared to the cropping practices under non-flooded conditions. It remains unclear, however, if and to what extent rice paddy management increases SOC storage in warmer climate regions where decomposition is much faster. We thus examined the factors controlling the SOC stock and the net change in SOC normalized to the duration under paddy (ΔSOC) at paddy fields in warm temperate (21 sites) and tropical climatic regions (11 sites) from peer-reviewed literature which was concentrated to Asia. Using collected data (e.g., SOC content, bulk density, mean annual temperature (MAT), precipitation, management duration, soil management, crop rotation), we evaluated predictor variables that account for ΔSOC variability (i.e., mean decrease accuracy) by random forest regression. The mean paddy duration for all data (n = 170) was 22 years. Average SOC stock and ΔSOC at 0–20 cm depth were 38 ± 1.45SE Mg C ha−1 and 0.34 ± 0.04SE Mg C ha−1 yr−1, respectively, indicating that continuous paddy management had a positive effect on SOC storage in both climatic regions. SOC stock was two-fold higher and ΔSOC was four-fold higher in the warm temperate region compared to the tropical region (p < 0.05). The best predictor variables for ΔSOC were MAT followed by soil management and then paddy duration. While organic amendment including both straw and manure increased ΔSOC in both warm temperate and tropical paddies, we did not detect a positive effect of chemical fertilizer application alone or crop rotation type (e.g., rice-upland vs rice-rice). Despite high variability, ΔSOC in the warm temperate region significantly declined over time from 0.82 ± 0.10 Mg C ha−1 yr−1 (< 20 yrs) to 0.22 ± 0.06 Mg C ha−1 yr−1. While SOC stock was 50% lower, the tropical paddies showed no clear decline in ΔSOC up to 41 years (0.04–0.21 Mg C ha−1 yr−1). These results suggest that SOC accretion might have reached equilibrium after 20 years of paddy practice in the warm temperate region and, if so, ΔSOC is likely to diminish to zero in the near future. In tropical paddy fields, on the other hand, slow SOC build-up may continue for a longer term although the ΔSOC value itself is not high. While the current study showed the efficacy of paddy management on SOC build-up under the warmer climate regions, further study on the rate and qualitative changes is necessary to evaluate the trade-offs between soil fertility improvement (e.g., soil N build-up) and methane emission.

Improvement of data imbalance for digital soil class mapping in Eastern China

A fine-resolution digital soil class map is needed. However, the problem of imbalanced data leads to an inaccurate spatial distribution of the digital soil class map, and the spatial resolution of digital soil class maps at a large scale is low in existing studies. Based on these points, an algorithm of over-sampling and under-sampling was introduced to solve the problem of imbalanced data, and to improve the performance of soil classification model. 316 topsoil samples with eight main soil classes at the great group level were collected in Eastern China. Eight out of twelve prediction variables were determined after the importance evaluation by “Mean Decrease Accuracy” in the random forest (RF) model, including digital elevation model (DEM), enhanced vegetation index (EVI), land surface wetness index (LSWI), land surface temperature (LST), normalized differenced vegetation index (NDVI), and soil texture components. RF model was also applied to complete digital soil class mapping, and the results of treated (over-sampling and under-sampling by randomly increasing or decreasing the number of samples) and untreated data were compared and discussed. Research results indicated that modeling by imbalanced data resulted in uncertain soil classes mapping, with minority classes were lost and with lower accuracies than those of balanced data (overall accuracy = 83.83 %, kappa coefficient = 0.79). After over-sampling and under-sampling treatments, these problems were well solved with an overall accuracy of 96.72 % and a kappa coefficient of 0.93. The accuracy of soil class prediction for minority classes were improved by 12.5 %–54.5 %. Compared to the existing conventional soil map, the new map with a fine resolution of 30 × 30 m is time-effective and more detailed. Validation (point-validation and map-to-map comparison) of the predicted map showed that the output is reliable and can provide a reference for other soil and environmental studies without major difficulties.

Use of Random Forest to Predict Adherence in an Online Intervention for Depression Using Baseline and Early Usage Data: Model Development and Validation on Retrospective Routine Care Log Data.

Online interventions, such as the iFightDepression (iFD) tool, are increasingly recognized as effective alternatives to traditional face-to-face psychotherapy or pharmacotherapy for treating depression. However, particularly when used outside of study settings, low adherence rates and the resulting diminished benefits of the intervention can limit their effectiveness. Understanding the factors that predict adherence would allow for early, tailored interventions for individuals at risk of nonadherence, thereby enhancing user engagement and optimizing therapeutic outcomes. This study aims to develop and evaluate a random forest model that predicts adherence to the iFD tool to identify users at risk of noncompletion. The model was based on characteristics collected during baseline and the first week of the intervention in patients with depression. Log data from 4187 adult patients who registered for the iFD tool between October 1, 2016, and May 5, 2022, and provided informed consent were statistically analyzed. The resulting data set was divided into training (2932/4187, 70%) and test (1255/4187, 30%) sets using a randomly stratified split. The training data set was utilized to train a random forest model aimed at predicting each user's adherence at baseline, based on the hypothesized predictors: age, self-reported gender, expectations of the intervention, current or previous depression treatments, confirmed diagnosis of depression, baseline 9-item Patient Health Questionnaire (PHQ-9) score, accompanying guide profession, and usage behavior within the first week. After training, the random forest model was evaluated on the test data set to assess its predictive performance. The importance of each variable in predicting adherence was analyzed using mean decrease accuracy, mean decrease Gini, and Shapley Additive Explanations values. Of the 4187 patients evaluated, 1019 (24.34%) were classified as adherent based on our predefined definition. An initial random forest model that relied solely on sociodemographic and clinical predictors collected at baseline did not yield a statistically significant adherence prediction. However, after incorporating each patient's usage behavior during the first week, we achieved a significant prediction of adherence (P<.001). Within this prediction, the model achieved an accuracy of 0.82 (95% CI 0.79-0.84), an F1-score of 0.53, an area under the curve of 0.83, and a specificity of 0.94 for predicting nonadherent users. The key predictors of adherence included logs, word count on the first workshop's worksheet, and time spent on the tool, all measured during the first week. Our results highlight that early engagement, particularly usage behavior during the first week of the online intervention, is a far greater predictor of adherence than any sociodemographic or clinical factors. Therefore, analyzing usage behavior within the first week and identifying nonadherers through the algorithm could be beneficial for tailoring interventions aimed at improving user adherence. This could include follow-up calls or face-to-face discussions, optimizing resource utilization in the process.