Random Forest Machine Learning Model Research Articles

Predictive Cultivation: Integrating Meteorological Data and Machine Learning for Enhanced Crop Yield Forecast

Agriculture is a key component of Telangana’s economy, and greater performance in this sector is crucial for inclusive growth. A central challenge is yielding estimation to predict crop yields before harvesting. This paper addresses this challenge with machine learning approaches includes Naive Bayes, KNN and Random Forest. The parameters considered for model testing are crop, season, rainfall and location. This paper includes a case study of Telangana with the help of Telangana weather data set to provide analysis on the key factors like overall rainfall recorded with respect to each Mandal, overall seasonal yield in selected years, seasonal yield of major crops like Bengal gram, groundnut and maize, and overall yield in two different agricultural seasons: rabi and kharif. Random forest machine learning model produces highest accuracy of 99.32% when compared with other process models.

Evaluation of the lithium resource in the Smackover Formation brines of southern Arkansas using machine learning.

Global demand for lithium, the primary component of lithium-ion batteries, greatly exceeds known supplies, and this imbalance is expected to increase as the world transitions away from fossil fuel energy sources. High concentrations of lithium in brines have been observed in the Smackover Formation in southern Arkansas (>400 milligrams per liter). We used published and newly collected brine lithium concentration data to train a random forest machine-learning model using geologic, geochemical, and temperature explanatory variables and create a map of predicted lithium concentrations in Smackover Formation brines across southern Arkansas. Using these predicted lithium maps with reservoir parameters and geologic information, we calculated that there are 5.1 to 19 million tons of lithium in Smackover Formation brines in southern Arkansas, which represents 35 to 136% of the current US lithium resource estimate. Based on these calculations, in 2022, 5000 tons of dissolved lithium were brought to the surface within brines as waste streams of the oil, gas, and bromine industries.

Hepatic Transcriptome Reveals Potential Key Genes Contributing to Differential Milk Production.

Despite the widespread adoption of TMR or PMR and the formulas designed to sufficiently cover the cows' requirements, individual dairy cows' milk production varies significantly. The liver is one of the most important organs in cow lactation metabolism and plays an essential role in the initiation of lactation. This study aimed to investigate the potential key genes in the liver contributing to the different milk production. We enrolled 64 cows and assigned them to high or low milk yield (MY) groups according to their first 3 weeks of milk production. We performed RNAseq for 35 liver samples with 18 from prepartum and 17 from postpartum cows. The continuous milk yield observation showed a persistently higher milk yield in high MY cows than low MY cows in the first 3 weeks. High MY cows showed better feed conversion efficiency. We identified 795 differentially expressed genes (DGEs) in the liver of high MY cows compared with low MY cows, with up-regulated genes linked to morphogenesis and development pathways. Weighted gene co-expression network analysis (WGCNA) revealed four gene modules positively correlating with milk yield, and protein and lactose yield (p < 0.05). Using the intersected genes between the four gene modules and DEGs, we constructed the linear mixed-effects models and identified six hub genes positively associated and two hub genes negatively associated with milk yield (Coefficients > 0.25, p < 0.05). Random forest machine learning model training based on these eight hub genes could efficiently predict the milk yield (p < 0.001, R2 = 0.946). Interestingly, the expression patterns of these eight hub genes remained remarkably similar before and after parturition. The present study indicated the critical role of liver in milk production. Activated processes involved in morphogenesis and development in liver may contribute to the higher milk production. Eight hub genes identified in this study may provide genetic research materials for dairy cow breeding.

Drought Quantification in Africa Using Remote Sensing, Gaussian Kernel, and Machine Learning

Effective drought management requires precise measurement, but this is challenging due to the variety of drought indices and indicators, each with unique methods and specific uses, and limited ground data availability. This study utilizes remote sensing data from 2001 to 2020 to compute drought indices categorized as meteorological, agricultural, and hydrological. A Gaussian kernel convolves these indices into a denoised, multi-band composite image. Further refinement with a Gaussian kernel enhances a single drought index from each category: Reconnaissance Drought Index (RDI), Soil Moisture Agricultural Drought Index (SMADI), and Streamflow Drought Index (SDI). The enhanced index, encompassing all bands, serves as a predictor for classification and regression tree (CART), support vector machine (SVM), and random forest (RF) machine learning models, further improving the three indices. CART demonstrated the highest accuracy and error minimization across all drought categories, with root mean square error (RMSE) and mean absolute error (MAE) values between 0 and 0.4. RF ranked second, while SVM, though less reliable, achieved values below 0.7. The results show persistent drought in the Sahel, North Africa, and southwestern Africa, with meteorological drought affecting 30% of Africa, agricultural drought affecting 22%, and hydrological drought affecting 21%.

Spot-checking machine learning algorithms for tool wear monitoring in automatic drilling operations in CFRP/Ti6Al4V/Al stacks in the aircraft industry

In aircraft manufacturing, where diverse materials, including Carbon Fiber-Reinforced Plastics (CFRP), aluminum, and titanium alloys, are employed, the assembly process heavily relies on creating thousands of holes. These holes accommodate bolts and rivets, facilitating the secure interlocking of structural components within the aircraft fuselage. The proliferation of sensor systems in this domain has led to a substantial increase in data generation during the hole-making process, offering a compelling opportunity to optimize the production system. In this context, this article is dedicated to harnessing the data collected from the production system of a commercial aircraft to refine the assembly process, with a specific focus on reducing consumable costs. The primary approach involves developing a real-time Tool Wear Monitoring System by comparing the performance of Linear Regression, Lasso Regression, Ridge Regression, k-Nearest Neighbors, Support Vector Regression, Decision Tree, Random Forest, and Extreme Gradient Boosting Machine Learning models. Using a scale of the general drill condition as an outcome, the Gradient Boosting Regressor has shown outstanding results. Notably, the residuals consistently exhibited zero-centered errors in training and test sets. However, it suggests that further enhancements are needed to surpass human-level performance in predicting tool conditions because of the quality and quantity of available data.

Predicting axillary lymph node metastasis in breast cancer based on ultrasound radiofrequency time-series analysis.

The status of axillary lymph nodes (ALN) plays a critical role in the management of patients with breast cancer. It is an urgent demand to develop highly accurate, non-invasive methods for predicting ALN status. To evaluate the efficacy of ultrasound radiofrequency (URF) time-series parameters, in combination with clinical data, in predicting ALN metastasis in patients with breast cancer. We prospectively gathered clinicopathologic and ultrasonic data from patients diagnosed with breast cancer. Various machine-learning (ML) models were developed using all available features to determine the most efficient diagnostic model. Subsequently, distinct prediction models were created using the optimal ML model, and their diagnostic performances were evaluated and compared. The study encompassed 240 patients, of whom 88 had lymph node metastases. A leave-one-out cross-validation (LOOCV) method was used to split the entire dataset into training and testing subsets. The random forest ML model outperformed the other algorithms, with an area under the curve (AUC) of 0.92. Prediction models based on clinical, ultrasonic, URF parameters, clinical + ultrasonic, clinical + URF, and ultrasonic + URF parameters had AUCs of 0.56, 0.79, 0.78, 0.90, 0.80, and 0.84, respectively, in the testing set. The comprehensive diagnostic model (clinical + ultrasonic + URF parameters) demonstrated strong diagnostic capability, with an AUC of 0.94 in the testing set, exceeding any single prediction model. The combined model (clinical + ultrasonic + URF parameters) could be used preoperatively to predict lymph node status, offering valuable input for the design of individualized surgical approaches.

A Comparative Study of Detrending Methods on Crop Yield Time Series for Drought Studies

Detrending is a statistical technique that removes systematic variations or trends from time series data, allowing analysts to focus on the underlying patterns or fluctuations. While multiple detrending approaches have been applied but rarely discussed their consistency of outcomes and effectiveness in accurately capturing better yield trends. The validation of drought occurrences has proven to be a challenging task due to the non-stationary characteristics of time series data related to crop yield. This research utilizes time series of cotton yield data from the Marathwada region covering the period from 1998 to 2021. Three traditional trend models, including simple linear regression, second-order polynomial regression and central moving average were applied. Additionally, two machine learning models (random forest and support vector regression) were tested with a novel approach. Moreover, two decomposition models (additive and multiplicative) were used to remove non-linear trends in crop yield time series data. The performance of the chosen models was evaluated based on metrics such as root mean square error, mean absolute error, Nash-Sutcliffe efficiency, and index of agreement. The results suggest that the most effective detrending approach involves combining a random forest machine learning model with an additive decomposition model.

Data-Driven Optimization of Plasma Electrolytic Oxidation (PEO) Coatings with Explainable Artificial Intelligence Insights

PEO constitutes a promising surface technology for the development of protective and functional ceramic coatings on lightweight alloys. Despite its interesting advantages, including enhanced wear and corrosion resistances and eco-friendliness, the industrial implementation of PEO technology is limited by its relatively high energy consumption. This study explores the development and optimization of novel PEO processes by means of machine learning (ML) to improve the coating thickness. For this purpose, ML models random forest and XGBoost were employed to predict the thickness of the developed PEO coatings based on the key process variables (frequency, current density, and electrolyte composition). The predictive performance was significantly improved by including the composition of the used electrolyte in the models. Furthermore, Shapley values identified the pulse frequency and the TiO2 concentration in the electrolyte as the most influential variables, with higher values leading to increased coating thickness. The residual analysis revealed a certain heteroscedasticity, which suggests the need for additional samples with high thickness to improve the accuracy of the model. This study reveals the potential of artificial intelligence (AI)-driven optimization in PEO processes, which could pave the way for more efficient and cost-effective industrial applications. The findings achieved further emphasize the significance of integrating interactions between variables, such as frequency and TiO2 concentration, into the design of processing operations.

Social determinants of health and outcome disparities in spine tumor surgery. Part 1: An analysis of 6.6 million nationwide admissions.

Earlier research has demonstrated that social determinants of health (SDoH) impact neurosurgical access and outcomes, but these trends are less characterized for spine tumors relative to intracranial tumors. The authors aimed to elucidate the association between SDoH and outcomes for a nationwide cohort of spine tumor surgery admissions. The authors identified all admissions with a spine tumor diagnosis in the National Inpatient Sample (NIS) from 2002 to 2019. Four SDoH were analyzed: race and ethnicity, insurance, household income, and safety-net hospital (SNH) treatment. Hospitals in the top quartile of safety-net burden (in terms of percentage of patients receiving Medicaid or uninsured) were categorized as SNHs. Multivariable regression queried the association between 22 variables and 5 perioperative outcomes: mortality, discharge disposition, complications, length of stay (LOS), and hospitalization costs. Interaction term analysis with hospitalization year was used to assess longitudinal changes in outcome disparities. Finally, the authors constructed random forest machine learning models to assess the impact of SDoH variables on prognostic accuracy and to quantify the relative importance of predictors for disposition. Of 6,593,392 total admissions with spine tumors, 219,380 (3.3%) underwent surgery. Non-White race (OR 0.80-0.91, p < 0.001) and nonprivate insurance (OR 0.76-0.83, p < 0.001) were associated with lower odds of receiving surgery. Among surgical admissions, presenting severity, including of myelopathy and plegia, was elevated among non-White, nonprivate insurance, and low-income admissions (all p < 0.001). Black race (OR 0.70, p < 0.001), Medicare (OR 0.70, p < 0.001), Medicaid (OR 0.90, p < 0.001), and lower income (OR 0.88-0.93, all p < 0.001) were associated with decreased odds of favorable discharge disposition. Increased LOS and costs were observed among non-White (+6%-10% in LOS and +5%-9% in costs, both p < 0.001) and Medicaid (+16% in LOS and +6% in costs, both p < 0.001) admissions. SNH treatment was also associated with higher mortality (OR 1.49, p < 0.001) and complication (OR 1.20, p < 0.001) rates. From 2002 to 2019, disposition improved annually for Medicaid patients (OR 1.03 per year, p = 0.022) but worsened for Black patients (OR 0.98 per year, p = 0.046). Random forest models identified household income as the most important predictor of discharge disposition. For spine tumor admissions, SDoH predicted surgical intervention, presenting severity, and perioperative outcomes. Over 2 decades, disparities improved for Medicaid patients but worsened for Black patients. Finally, SDoH significantly improve prognostic accuracy for outcomes after spine tumor surgery. Further study toward ameliorating patient disparities for this population is warranted.

Synergistic patient factors are driving recent increased pediatric urgent care demand.

We aimed to use the high fidelity urgent care patient data to model the factors that have led to the increased demand at our local pediatric urgent care centre. The dataset for this retrospective cohort study was obtained from our local healthcare centre's national reporting data for pediatric urgent care visits from 2006 to 2022. Variables analyzed included: basic patient demographics, chief complaint, triage urgency, date and time of registration/discharge, discharge diagnosis, and discharge destination. Statistical analysis of non-linear trends was summarized by locally estimated scatterplot smoothing splines. For machine learning, we used the tidymodels R package. Models were validated in training using k-fold cross validation with k = 5. We used univariate linear regression as a baseline model. After the data was standardized, correlation and homoscedasticity were evaluated between all parameter permutations. This dataset consisted of 164,660 unique visits to our academic centre's pediatric urgent care. Over the study period, there was an overall substantial increase in the number of urgent care visits per day, with a rapid increase beyond previous levels in 2021 and further in 2022. The increased length of stay trend was consistent across presenting complaint categories. The proportion of patients without primary care in 2022 was 2.5 times higher than in 2013. A random forest machine learning model revealed the relative importance of features to predicting a visit in 2022 were: longer stay, later registration in the day, diagnosis of an infectious illness, and younger age. This study identified a combination of declining primary care access, circulating viral infections, and shifting chief complaints as factors driving the recent increase in frequency and duration of visits to our urgent care service.

Integrated finite element analysis and machine learning approach for propagation pressure prediction in hybrid Steel-CFRP subsea pipelines

Accurate prediction of the propagation pressure (PP) in hybrid steel-CFRP pipe systems presents a substantial challenge due to intricate interactions and complex collapse failure modes. An efficient FE-based algorithm is programmed using ANSYS to numerically estimate the PP of hybrid steel-CFRP pipe, subjected to external pressure. This study employs a machine learning (ML) framework, addressing the inherent complexity with a three-phase approach: Parameter Design, Buckle Propagation Analysis, and ML Model Development. The dataset, encompassing about two thousand observations with four key features, undergoes k-fold cross-validation and min-max normalization for robust ML performance. Five ML models—Random Forest (RF), K-Nearest Neighbors (KNN), Genetic Programming (GP), Multi-layer Perceptron (MLP), and Support Vector Machine (SVM)—are developed and evaluated. The results revealed a significant influence of Ds/ts, a three-phase relationship with ts/tc, and a substantial decrease in PPh/PPs with increasing σys/σuc, predominantly exhibiting U-shaped or dog-bone failure modes in different scenarios. Proven that GP, KNN, and RF are the superior performers, ranking ahead of SVM with Gaussian Kernel (SVM-GK), MLP, and SVM with Linear Kernel (SVM-LK). Statistical metrics, Taylor Diagram analysis, and comparisons with FE results emphasize the effectiveness of GP, KNN, and RF. Additionally, normality tests and feature importance analysis provide nuanced insights.

L Test Subtask Segmentation for Lower-Limb Amputees Using a Random Forest Algorithm.

Functional mobility tests, such as the L test of functional mobility, are recommended to provide clinicians with information regarding the mobility progress of lower-limb amputees. Smartphone inertial sensors have been used to perform subtask segmentation on functional mobility tests, providing further clinically useful measures such as fall risk. However, L test subtask segmentation rule-based algorithms developed for able-bodied individuals have not produced sufficiently acceptable results when tested with lower-limb amputee data. In this paper, a random forest machine learning model was trained to segment subtasks of the L test for application to lower-limb amputees. The model was trained with 105 trials completed by able-bodied participants and 25 trials completed by lower-limb amputee participants and tested using a leave-one-out method with lower-limb amputees. This algorithm successfully classified subtasks within a one-foot strike for most lower-limb amputee participants. The algorithm produced acceptable results to enhance clinician understanding of a person's mobility status (>85% accuracy, >75% sensitivity, >95% specificity).

Machine learning-derived prognostic signature for progression-free survival in non-metastatic nasopharyngeal carcinoma.

Early detection of high-risk nasopharyngeal carcinoma (NPC) recurrence is essential. We created a machine learning-derived prognostic signature (MLDPS) by combining three machine learning (ML) models to predict progression-free survival (PFS) in patients with non-metastatic NPC. A cohort of 653 patients with non-metastatic NPC was divided into a training (n = 457) and validation (n = 196) dataset (7:3 ratio). The study included clinicopathological characteristics, hematologic markers, and MRI findings in three machine learning models-random forest (RF), extreme gradient boosting (XGBoost), and least absolute shrinkage and selection operator (LASSO)-to predict progression-free survival (PFS). A Venn diagram identified the overlapping signatures from the three ML algorithms. Cox proportional hazard analysis determined the MLDPS for PFS. The RF, XGBoost, and LASSO algorithms identified six consensus factors from the 33 signatures. Cox proportional hazards analysis showed that the MLDPS includes age, lymphocyte count, number of positive lymph nodes, and regional lymph node density. Additionally, MLDPS effectively stratified prognosis, with low-risk individuals showing better PFS than high-risk individuals (p < 0.001). MLDPS, based on clinicopathological characteristics, hematologic markers, and MRI findings, is crucial for guiding clinical management and personalizing treatments for patients with non-metastatic NPC.

Analyses of groundwater level in a data-scarce region based on assessed precipitation products and machine learning

Groundwater, a vital natural resource globally, faces challenges related to data scarcity, particularly in data-limited regions. To address this issue in the region of Bahira in Morocco, we developed an innovative approach for estimating groundwater levels by utilizing precipitation products and employed a random forest (RF) machine learning (ML) model to fill data gaps. Our study involved a comprehensive assessment of ten precipitation products, comprising six Reanalysis Precipitation Products (RPPs) and four Satellite-based Precipitation Products (SPPs). We evaluated their performance across various time scales, including daily, monthly, and seasonal, across different topographical classes. The outcomes highlighted the consistency of ERA5-based datasets, boasting daily correlation to values higher than 0.6, whereas monthly and seasonal correlations exceed 0.8, except during summer. GPM-IMERG, MERRA2, and CPC-UPP also demonstrated commendable accuracy, particularly in plain and mountainous areas. Nonetheless, CFSR, CHIRPS, PERSIANN CDR, and TRMM datasets exhibited limitations, particularly in high mountain areas. To address data gaps, we initially explored correlations between RPPs, SPPs, and groundwater data. However, these correlations failed to meet the accuracy standards required for precise predictions. Notably, the strongest correlations were observed in monitoring stations located in mountainous regions, indicating significant aquifer recharge activities in these areas. In the subsequent phase, the Multiple Imputation by Chained Equations (MICE) machine learning-based imputation method served as a valuable tool for estimating groundwater levels in regions where ground observations were insufficient. Our trend analysis yielded significant insights, with approximately 95% of groundwater points displaying negative trends, with a maximum rate of −0.91 m. In contrast, 69% of precipitation stations exhibited negative trends, with a maximum rate of −0.06 mm. Our approach offers a promising potential to address the challenges associated with the scarcity of groundwater and precipitation data, making it a valuable tool for the assessment, monitoring, and management of groundwater resources.

Research on credit risk assessment optimization based on machine learning

Credit business is a vital part of the bank's core business, which has an extremely important impact on the bank's income and development. In the operation of credit business, credit risk assessment is particularly crucial, and accurate risk assessment can minimize risks while maximizing the bank's returns. We propose a method to optimize credit risk assessment using machine learning techniques. In this work, we employ a random forest machine learning model to process and analyze large amounts of loan application data. By using correlation analysis, information enrichment, etc., the characteristics that have the most impact on credit risk assessment are screened. Subsequently, the model was constructed using a random forest algorithm. Random forests improve the generalization ability and accuracy of the model by building multiple decision trees and introducing randomness between these trees. In the experimental analysis part, we compare the performance of various models on the German credit dataset, and the results show that the deep learning model outperforms the traditional machine learning model in most indicators, verifying the effectiveness of our method.

Feasibility of virtual reality and machine learning to assess personality traits in an organizational environment.

Personality plays a crucial role in shaping an individual's interactions with the world. The Big Five personality traits are widely used frameworks that help describe people's psychological behaviours. These traits predict how individuals behave within an organizational setting. In this article, we introduce a virtual reality (VR) strategy for relatively scoring an individual's personality to evaluate the feasibility of predicting personality traits from implicit measures captured from users interacting in VR simulations of different organizational situations. Specifically, eye-tracking and decision-making patterns were used to classify individuals according to their level in each of the Big Five dimensions using statistical machine learning (ML) methods. The virtual environment was designed using an evidence-centered design approach. The dimensions were assessed using NEO-FFI inventory. A random forest ML model provided 83% accuracy in predicting agreeableness. A k-nearest neighbour ML model provided 75%, 75%, and 77% accuracy in predicting openness, neuroticism, and conscientiousness, respectively. A support vector machine model provided 85% accuracy for predicting extraversion. These analyses indicated that the dimensions could be differentiated by eye-gaze patterns and behaviours during immersive VR. Eye-tracking measures contributed more significantly to this differentiation than the behavioural metrics. Currently, we have obtained promising results with our group of participants, but to ensure the robustness and generalizability of our findings, it is imperative to replicate the study with a considerably larger sample. This study demonstrates the potential of VR and ML to recognize personality traits.

A random forest machine learning model to detect fluvial hazards

AbstractFluvial hazards of river mobility and flooding are often problematic for road infrastructure and need to be considered in the planning process. The extent of river and road infrastructure networks and their tendency to be close to each other creates a need to be able to identify the most dangerous areas quickly and cost‐effectively. In this study, we propose a novel methodology using random forest (RF) machine learning methods to provide easily interpretable fine‐scale fluvial hazard predictions for large river systems. The tools developed provide predictions for three models: presence of flooding (PFM), presence of mobility (PMM) and type of erosion model (TEM, lateral migration, or incision) at reference points every 100 m along the fluvial network of three watersheds within the province of Quebec, Canada. The RF models use variables focused on river conditions and hydrogeomorphological processes such as confinement, sinuosity, and upstream slope. Training/validation data included field observations, results from hydraulic and erosion models, government infrastructure databases, and hydro‐ geomorphological assessments using 1‐m DEM and satellite/historical imagery. A total of 1807 reference points were classified for flooding, 1542 for mobility, and 847 for the type of erosion out of the 11,452 reference points for the 1145 km of rivers included in the study. These were divided into training (75%) and validation (25%) datasets, with the training dataset used to train supervised RF models. The validation dataset indicated the models were capable of accurately predicting the potential for fluvial hazards to occur, with precision results for the three models ranging from 83% to 94% of points accurately predicted. The results of this study suggest that RF models are a cost‐effective tool to quickly evaluate the potential for fluvial hazards to occur at the watershed scale.

Wet inrush susceptibility assessment at the Deep Ore Zone mine using a random forest machine learning model

In cave mines, wet inrushes occur when there is an uncontrolled inflow of fine, wet material from drawpoints. Currently, uncertainty exists regarding the spatial-temporal pattern and severity of inrush incidents. This uncertainty arises from the limited understanding of wet inrush mechanisms within the complex conditions of a cave mine. In this study, the existing gaps in knowledge around the spatial and temporal patterns of inrush incidents were addressed using machine learning techniques. A random forest (RF) model was employed to analyse the inrush database collected at the Deep Ore Zone mine over several years. The conceptual understanding of inrush mechanisms and triggers, along with historical evidence, was employed to establish an initial set of key inrush variables to be used in the RF model. The developed RF model demonstrated promising performance with an accuracy of 85%. The feature importance results indicated that previous inrush history, fragment size, draw rate (short term and long term), differential draw index (short term and long term) and history of inrush at neighbouring drawpoints had the highest impact on inrush susceptibility. The insights gained provide an improved assessment of inrush susceptibility, thereby improving the strategies employed to mitigate inrush risk.

Serum metabolomic profile of myasthenia gravis and potential values as biomarkers in disease monitoring

ObjectiveSerum metabolites from 19 myasthenia gravis (MG) patients and 15 normal controls were analyzed via untargeted metabolomics, including 6 pre/post-treatment paired MG patients, to assess the value of serum metabolites as biomarkers in monitoring MG. MethodDifferential metabolites between MG patients and normal controls were identified through liquid and gas chromatography-mass spectrometry simultaneously. Principal component analysis and orthogonal partial least squares-discriminant analysis were conducted to identify the differential metabolites. Candidate metabolites and pathways associated with MG were selected through a random forest machine learning model. ResultA total of 310 differential metabolites were identified with a threshold of variable projected importance > 1 and P value < 0.05. Among these, 158 metabolites were upregulated and 152 were downregulated. The random forest machine learning model selected 5 metabolites as potential biomarkers associated with MG: lignoceric acid (AUC=0.944), uridine diphosphate-N-acetylglucosamine (AUC=0.951), arachidonic acid (AUC=0.951), beta-glycerophosphoric acid (AUC=0.933), and L-Asparagine (AUC=0.877). Further analysis using 6 paired MG patients pre- and post-immunosuppression treatment revealed 25 upregulated and 6 downregulated metabolites in post-treatment serum, which might be relevant to disease intervention. The significance remains elusive due to the limited number of patients. ConclusionA subset of differential metabolites was identified in the serum of MG patients, some of which changed with immunosuppressive therapy. Small molecule metabolites may serve as valuable biomarkers for disease monitoring in MG.

Toxic effects of trace metal(loid) mixtures on aquatic organisms

The co-occurrence of metal (loid)s in realistic aquatic environments necessitates the evaluation of their combined effects. However, the generality of the additive effect hypothesis is contentious, particularly due to metal(loid)-metal(loid) interactions. The absence of systematic evaluation approaches restricts our ability to draw overall conclusions and make reliable predictions. In this study, we reviewed 1473 effect sizes from 38 publications, and classified all responses into seven main categories (from molecular to individual levels) according to their toxicological significance. Our meta-analysis revealed that metal(loid) mixtures had significant effects on aquatic organisms (33 %, 95 % CI 28 %–39 %, P < 0.05), along with significant response heterogeneity (Qt = 690,319.62, P < 0.0001; I2 = 99.95 %). Concurrently, we developed a Random Forest machine learning model to predict adverse effects and identify key variables. These two methods demonstrated that the toxicity of metal(loid) mixtures is primarily linked to the choice of toxicity endpoints, and the characteristics of metal(loid) mixtures. Our findings underscore the potential of combining meta-analysis with machine learning, a more systematic approach, to enhance the understanding and prediction of the adverse effects of metal(loid) mixtures, and they offer guidance for risk assessment and policy-making in complex environmental scenarios.