Ensemble Learning Techniques Research Articles

CT radiomics-based biomarkers can predict response to immunotherapy in hepatocellular carcinoma

Hepatocellular Carcinoma (HCC) remains a leading cause of cancer deaths. Despite the rise of immunotherapies, many HCC patients don't benefit. There's a clear need for biomarkers to guide treatment decisions. This research aims to identify such biomarkers by combining radiological data and machine learning. We analyzed clinical and CT imaging data of 54 HCC patients undergoing immunotherapy. Radiologic features were examined to develop a model predicting short-term immunotherapy effects. We utilized 9 machine learning and 2 ensemble learning techniques using RapidMiner for model construction. We conducted the validation of the above feature combination using 29 HCC patients who received immunotherapy from another hospital, and tested and validated it using XGBoost combined with a genetic algorithm. In 54 HCC patients, radiomics features varied significantly between those with partial response (PR) and stable disease (SD). Key features in Gray Level Run Length Matrix (GLRLM) and in adjacent tissues' Intensity Direct, Neighborhood Gray Tone Difference Matrix (NGTDM), and Shape correlated with short-term immunotherapy efficacy. Selected feature combinations of 15, 19, and 8/15 features yielded better outcomes. Deep learning, random forest, and naive bayes outperformed other methods, with Bagging being more reliable than Adaboost. In the validation set of 29 HCC patients, the mentioned feature combination also demonstrated favorable performance. Furthermore, we achieved improved results when employing XGBoost in conjunction with a genetic algorithm for testing and validation. The machine learning model built with CT image features derived from GLCM, GLRLM, IntensityDirect, NGTDM, and Shape can accurately forecast the short-term efficacy of immunotherapy for HCC.

Prediction and Analysis of Rice Production and Yields Using Ensemble Learning Techniques

This research focuses on predicting and analyzing rice production and yield throughout the world using ensemble learning techniques. The study applies and compares three methods: linear regression, ARIMA, and ensemble learning, to predict rice harvest yields. The results show that ensemble learning techniques significantly improve prediction performance. For instance, the ensemble model for predicting area harvested, combining Model 6 (linear regression) and Model 10 (ARIMA), achieved of coefficient of determination outperforming the individual models. Similarly, for predicting yield, the ensemble model combining Model 4 (linear regression) and Model 9 (ARIMA) achieved of coefficient of determination indicating superior prediction accuracy. For predicting production, the ensemble model combining Model 2 (linear regression) and Model 8 (ARIMA) achieved of coefficient of determination. These results demonstrate the effectiveness of ensemble learning in enhancing prediction accuracy with lower MSE and RMSE values. By analyzing various factors influencing rice yields, this research provides valuable insights for increasing rice production and yield, supporting efforts to improve the efficiency and effectiveness of rice farming, and contributing to achieving the United Nations Sustainable Development Goals (SDGs).

A machine learning approach for predicting flexural strength of 3D printed hexagon lattice-cored sandwich structures

The development of sandwich structures using Fused Filament Fabrication (FFF) technology is an effective method for the rapid construction of complex profiles and is widely recognized for various structural applications. In this study, hexagon lattice-cored sandwich structures are developed by placing the lattice core in the center portion of the PLA polymeric specimen. The performance is analyzed with respect to varying levels of 3D-Printing Factors (3D-PFs) such as Nozzle Temperature (NT), Line Width (LW), Printing Speed (PS), and Layer Height (LH). The levels of the respective 3D-PFs are varied as follows: NT (180, 190, 200, and 210 °C), LH (0.15, 0.2, 0.25, and 0.3 mm), PS (15, 20, 25, and 30 mm/s), and LW (0.1, 0.2, 0.3, and 0.4 mm). The sandwich flexural samples are 3D printed using an FFF 3D printer, and their flexural properties are examined using a Universal Testing Machine (UTM). In the present work, various Machine Learning (ML) algorithms, such as Least Angle Regression (LARS), Support Vector Machine (SVM), Decision Tree (DT), and Ridge Regression (Ridge), are employed to predict the Flexural Strength (FS) of the 3D-printed sandwich structures and help in determining the optimized levels of 3D-PFs for achieving higher FS. The findings indicate that the LARS algorithm, particularly when applied within the Bagging Ensemble Learning Technique (ELT), exhibits the highest precision, showcasing a Mean Absolute Error (MAE) of 0.163, Root Mean Square Error (RMSE) of 0.153, and Median Absolute Error (MedAE) of 0.129. With the help of the LARS algorithm under Bagging ELT, the impact of 3D-PFs on the resulting FS is analyzed, which helps in determining the optimized levels of 3D-PFs concerning FS. The sandwich structures fabricated at the optimized levels show improved flexural properties, making them suitable for various structural applications.

Implementation of Ensemble Learning Algorithm - Stacking Regressor on PM2.5 Prediction Model

This study aims to develop a PM2.5 concentration prediction model using ensemble learning techniques, focusing on the application of stacking regressor. The developed model is compared with several other basic models, namely LSTM, Random Forest, XGBoost, and GBM, to evaluate its performance in terms of prediction accuracy. The results show that the stacking regressor model provides more accurate prediction results than these basic models. The LSTM model has a Mean Squared Error (MSE) value of 80.42 and a coefficient of determination (R²) of -0.29, which shows unsatisfactory performance despite being able to capture temporal patterns. Random Forest gave better results with an MSE of 10.90 and R² of 0.83, thanks to its ability to handle heterogeneous data. The XGBoost and GBM models showed similar performance, with MSE of 9.01 and 8.81 respectively and R² of 0.86. However, the stacking regressor model with the RidgeCV meta-learner achieved the best results with an MSE of 8.07 and R² of 0.87, indicating that this technique successfully combines the advantages of various base models to improve prediction accuracy. In conclusion, the stacking regressor model proved effective in improving PM2.5 prediction accuracy, making it a potential tool for air quality monitoring and supporting public health policy making. This research makes an important contribution to the development of ensemble learning-based air quality prediction methods, and the results can serve as a reference for future research

Interval-Censored Linear Quantile Regression

Censored quantile regression has emerged as a prominent alternative to classical Cox’s proportional hazards model or accelerated failure time model in both theoretical and applied statistics. While quantile regression has been extensively studied for right-censored survival data, methodologies for analyzing interval-censored data remain limited in the survival analysis literature. This article introduces a novel local weighting approach for estimating linear censored quantile regression, specifically tailored to handle diverse forms of interval-censored survival data. The estimation equation and the corresponding convex objective function for the regression parameter can be constructed as a weighted average of quantile loss contributions at two interval endpoints. The weighting components are nonparametrically estimated using local kernel smoothing or ensemble machine learning techniques. To estimate the nonparametric distribution mass for interval-censored data, a modified EM algorithm for nonparametric maximum likelihood estimation is employed by introducing subject-specific latent Poisson variables. The proposed method’s empirical performance is demonstrated through extensive simulation studies and real data analyses of two HIV/AIDS datasets. Supplementary materials for this article are available online.

Detection of Subtle ECG Changes Despite Superimposed Artifacts by Different Machine Learning Algorithms

Analyzing electrocardiographic (ECG) signals is crucial for evaluating heart function and diagnosing cardiac pathology. Traditional methods for detecting ECG changes often rely on offline analysis or subjective visual inspection, which may overlook subtle variations, particularly in the case of artifacts. In this theoretical, proof-of-concept study, we investigated the potential of five different machine learning algorithms [random forests (RFs), gradient boosting methods (GBMs), deep neural networks (DNNs), an ensemble learning technique, as well as logistic regression] to detect subtle changes in the morphology of synthetically generated ECG beats despite artifacts. Following the generation of a synthetic ECG beat using the standardized McSharry algorithm, the baseline ECG signal was modified by changing the amplitude of different ECG components by 0.01–0.06 mV. In addition, a Gaussian jitter of 0.1–0.3 mV was overlaid to simulate artifacts. Five different machine learning algorithms were then applied to detect differences between the modified ECG beats. The highest discriminatory potency, as assessed by the discriminatory accuracy, was achieved by RFs and GBMs (accuracy of up to 1.0), whereas the least accurate results were obtained by logistic regression (accuracy approximately 10% less). In a second step, a feature importance algorithm (Boruta) was used to discriminate which signal parts were responsible for difference detection. For all comparisons, only signal components that had been modified in advance were used for discretion, demonstrating that the RF model focused on the appropriate signal elements. Our findings highlight the potential of RFs and GBMs as valuable tools for detecting subtle ECG changes despite artifacts, with implications for enhancing clinical diagnosis and monitoring. Further studies are needed to validate our findings with clinical data.

A Novel Model Based on CNN-ViT Fusion and Ensemble Learning for the Automatic Detection of Pes Planus.

Background: Pes planus, commonly known as flatfoot, is a condition in which the medial arch of the foot is abnormally low or absent, leading to the inner part of the foot having less curvature than normal. Symptom recognition and errors in diagnosis are problems encountered in daily practice. Therefore, it is important to improve how a diagnosis is made. With the availability of large datasets, deep neural networks have shown promising capabilities in recognizing foot structures and accurately identifying pes planus. Methods: In this study, we developed a novel fusion model by combining the Vgg16 convolutional neural network (CNN) model with the vision transformer ViT-B/16 to enhance the detection of pes planus. This fusion model leverages the strengths of both the CNN and ViT architectures, resulting in improved performance compared to that in reports in the literature. Additionally, ensemble learning techniques were employed to ensure the robustness of the model. Results: Through a 10-fold cross-validation, the model demonstrated high sensitivity, specificity, and F1 score values of 97.4%, 96.4%, and 96.8%, respectively. These results highlight the effectiveness of the proposed model in quickly and accurately diagnosing pes planus, making it suitable for deployment in clinics or healthcare centers. Conclusions: By facilitating early diagnosis, the model can contribute to the better management of treatment processes, ultimately leading to an improved quality of life for patients.

Design of a precise ensemble expert system for crop yield prediction using machine learning analytics

AbstractAgriculture is facing significant challenges in the development of crop yield forecasts, which are important aspects of decision‐making at the international, regional, and local levels. The area of agriculture is attracting growing attention because of increasing the demand for food supplies. To ensure future food supplies, crop yield prediction (CYP) provides the best decision‐making to assist farmers in agricultural yield forecasting efficiently. Nevertheless, CYP is a difficult endeavor because of the intricacy of the underlying mechanisms and the effect of numerous factors, including weather patterns, soil characteristics, and crop management techniques. In today's era, ensemble learning (EL) approaches have recently demonstrated significant promise for enhancing the reliability and accuracy of CYP. The success of the EL techniques depends on several facts, including how the base learner models are trained and how these are combined. This study provides important insights into the EL techniques for CYP. This paper proposes an expert system model named precise ensemble expert system for crop yield prediction (PEESCYP) to predict the best crop for agricultural land. The proposed PEESCYP model employs multiple imputation by chained equation (MICE) data imputation technique to treat the missing values of the collected dataset, the isolation forest (IF) technique for outlier detection, the ant colony optimization (ACO) technique to perform feature selection, robust scaling (RS) technique to perform data normalization, and the extra tree (ET) is used for classification to overcome the variance and overfitting problem of the single classifiers. The measurements of the proposed PEESCYP model have been collected by means of accuracy, precision, recall, and F‐1 score using a prepared dataset, which is collected from International Crops Research Institute for the Semi‐Arid Tropics (ICRISAT), and the proposed model is compared with different single‐classifier based ML models, EL models, and various existing models available in the literature. The results of this experiment underline that the proposed PEESCYP model outperforms the others.

High-Resolution Rainfall Estimation Using Ensemble Learning Techniques and Multisensor Data Integration.

In Indonesia, the monitoring of rainfall requires an estimation system with a high resolution and wide spatial coverage because of the complexities of the rainfall patterns. This study built a rainfall estimation model for Indonesia through the integration of data from various instruments, namely, rain gauges, weather radars, and weather satellites. An ensemble learning technique, specifically, extreme gradient boosting (XGBoost), was applied to overcome the sparse data due to the limited number of rain gauge points, limited weather radar coverage, and imbalanced rain data. The model includes bias correction of the satellite data to increase the estimation accuracy. In addition, the data from several weather radars installed in Indonesia were also combined. This research handled rainfall estimates in various rain patterns in Indonesia, such as seasonal, equatorial, and local patterns, with a high temporal resolution, close to real time. The validation was carried out at six points, namely, Bandar Lampung, Banjarmasin, Pontianak, Deli Serdang, Gorontalo, and Biak. The research results show good estimation accuracy, with respective values of 0.89, 0.91, 0.89, 0.9, 0.92, and 0.9, and root mean square error (RMSE) values of 2.75 mm/h, 2.57 mm/h, 3.08 mm/h, 2.64 mm/h, 1.85 mm/h, and 2.48 mm/h. Our research highlights the potential of this model to accurately capture diverse rainfall patterns in Indonesia at high spatial and temporal scales.

Ensemble learning techniques against structured query language injection attacks

Structured query language (SQL) injection threats pose severe risks to web applications, necessitating robust detection measures. This study introduced DSQLIA, employing ensemble learning algorithms-Bagging, Stacking, and AdaBoost classifiers-for SQL injection detection. Results unveiled the bagging classifier's 84% accuracy with perfect precision (100%) but moderate recall (68%). The stacking classifier achieved 85% accuracy, exceptional precision (99%), and balanced memory (72%), yielding an 83% F1-Score. Remarkably, the AdaBoost classifier outperformed, achieving 99% accuracy, high precision (98%), and outstanding recall (99%), leading to a remarkable 99% F1-Score. These findings highlight AdaBoost's superior ability to identify malicious queries with minimal false positives accurately. Overall, this research underscores the potential of ensemble learning in fortifying web application security against SQL injection attacks, emphasizing the AdaBoost classifier's exceptional performance in achieving precise and comprehensive detection.

Tunnel construction worker safety state prediction and management system based on AHP and anomaly detection algorithm model

Tunnels represent complex, high-risk, and technically demanding underground construction projects. The safety of construction workers in tunnels is influenced by various factors, including physiological indicators, tunnel dimensions, and internal environmental conditions. Analyzing safety based solely on static factors is inadequate for modern tunnel engineering safety management requirements. To address this challenge, this paper provides a comprehensive analysis of factors impacting safety and employs the Analytic Hierarchy Process (AHP) to identify seven significant factors with high importance: body temperature, heart rate, internal temperature, internal humidity, CO concentration, chlorine concentration, and the relative positioning of personnel. Considering these factors essential for assessing worker safety, we introduce a novel model named Tunnel-APH-AD. For training models aimed at anomaly detection, we performed data augmentation and utilized four distinct machine learning models. Additionally, ensemble learning techniques were applied to aggregate the predictions from individual models, thereby enhancing the effectiveness of detecting safety states for tunnel workers. We also evaluated the performance of these models on out-of-distribution (OOD) samples to test their robustness and generalizability. The experimental results indicate that, under similar ventilation and tunnel conditions, the ensemble learning model exhibits superior overall performance compared to individual models, underscoring the effectiveness of model combination in improving the accuracy and reliability of safety alerts. Through experimental validation, this study provides interpretable, scalable, and scientifically generalized applications of machine learning theories in systems for tunnel construction worker safety alerts. These findings contribute to advancing safety management practices in tunnel engineering, enabling proactive and effective measures to mitigate potential risks and ensure the well-being of workers.

Research on investment project risk prediction and management based on machine learning

In the era of digital economy, digital transformation is an inevitable choice in line with current economic development and national policy trends. Enterprises use new generation digital technologies such as big data, blockchain, cloud computing, artificial intelligence, and financial technology to apply these technologies to various production and business activities. The research on project risk management has gradually introduced the method of intelligent decision-making, using technologies such as big data and artificial intelligence to identify and analyze risks. We use learning models and ensemble learning techniques to predict and manage the risks of investment projects. Through long short-term memory networks (LSTMs), we are able to effectively extract spatiotemporal features from historical investment data to predict future risk dynamics. In order to further improve the accuracy of prediction and the robustness of the model, we introduced the Gradient Booster (GBM) ensemble learning method. These integrated technologies not only optimize the overall performance of the model, but also enhance the adaptability and forecast accuracy of various market changes. In the experimental analysis, we compare the performance of multiple models on real-world investment datasets, and the results show that the ensemble learning method has significant advantages in risk prediction accuracy.

INTEGRATING NEURAL NETWORKS INTO SHEET METAL FORMING: A REVIEW OF RECENT ADVANCES AND APPLICATIONS

In order to predict defects, improve performance, and streamline operations, machine learning techniques are becoming ever more indispensable in manufacturing processes, mainly in sheet metal forming. Incorporating neural networks into the process of sheet metal forming is the subject of this article's exhaustive examination of recent developments and applications. Exploring datasets from a variety of sheet metal forming processes, numerous machine learning models, including ensemble and single learning techniques are investigated. The functionality of this method extends to various tasks, including the prediction of springback in cold-rolled anisotropic steel sheets. The review provides a conclusion section that presents the main implementation methodologies and how they address to some manufacturing issues.

Enhance the Detection of DoS and Brute Force Attacks within the MQTT Environment Through Feature Engineering and Employing an Ensemble Technique

The rapid development of the Internet of Things (IoT) environment has introduced unprecedented levels of connectivity and automation. The Message Queuing Telemetry Transport (MQTT) protocol has become recognized in IoT applications due to its lightweight and efficient features; however, this simplicity also renders MQTT vulnerable to multiple attacks that can be launched against the protocol, including denial of service (DoS) and brute-force attacks. This study aims to improve the detection of intrusion DoS and bruteforce attacks in an MQTT traffic intrusion detection system (IDS). Our approach utilizes the MQTT dataset for model training by employing effective feature engineering and ensemble learning techniques. Following our analysis and comparison, we identified the top 10 features demonstrating the highest effectiveness, leading to improved model accuracy. We used supervised machine learning models, including Random Forest, Decision Trees, k-Nearest Neighbors, and XGBoost, in combination with ensemble classifiers. Stacking, voting, and bagging ensembles utilize these four supervised machinelearning methods to combine models. This study's results illustrate the proposed technique's efficacy in enhancing the accuracy of detecting DoS and brute-force attacks in MQTT traffic. Stacking and voting classifiers achieved the highest accuracy of 0.9538. Our approach outperforms the most recent study that utilized the same dataset.

Predicting grout’s uniaxial compressive strength (UCS) for fully grouted rock bolting system by applying ensemble machine learning techniques

This study proposes a novel system for accurately predicting grout’s uniaxial compressive strength (UCS) in fully grouted rock bolting systems. To achieve this, a database comprising 73 UCS values with varying water-to-grout (W/G) ratios ranging from 22 to 42%, curing times from 1 to 28 days, the admixture of fly ash contents ranging from 0 to 30%, and two Australian commercial grouts, Stratabinder HS, and BU-100, was built after conducting comprehensive series of experimental tests. After building the dataset, a metaheuristic technique, the jellyfish search (JS) algorithm was employed to determine the weight of base models in the ensemble system. This system combined various data and modelling techniques to enhance the accuracy of the UCS predictions. What sets this technique apart is the comprehensive database and the innovative use of the JS algorithm to create a weighted averaging ensemble model, going beyond traditional methods for predicting grout strength. The proposed ensemble model was called the weighted averaging ensemble model (WAE-JS), in which the obtained results of several soft computing models such as multi-layer perceptron (MLP), Bayesian regularized (BR) neural networks, generalized feed-forward (GFF) neural networks, classification and regression tree (CART), and random forest (RF) were weighted based on JS and the new results were then generated. Eventually, the result of WAE-JS was compared to other models, including MLP, BR, GFF, CART, and RF, based on some statistical parameters, such as R-squared coefficients, RMSE, and VAF as indices for evaluating the performance and capability of the proposed model. The results suggested the superiority of the ensemble WAE-JS system over the base models. In addition, the proposed WAE-JS model effectively improved the predicting accuracy achieved from the MLP, BR, GFF, CART, and RF. Furthermore, the sensitivity analysis revealed that the W/G had the most significant impact on the grout’s UCS values.

Profiling of particulate matter transport flux based on dual-wavelength lidar and ensemble learning algorithm.

Transport flux (TF) is a significant particulate matter (PM) characteristic. This paper introduces an advanced dual-wavelength polarization aerosol and wind lidar (Wind Flux 3000) capable of independently observing the PM TF. The PM TF observation capability, which allows for simultaneous aerosol and wind measurements, was achieved by integrating a Mie-polarization particle lidar module and a coherent Doppler wind lidar module into a single lidar system. The primary measurement products of the Wind Flux 3000 include particulate extinction coefficient at 532 nm and 1550 nm, volume linear depolarization ratio at 532 nm (δp,532), wind speed (WS), wind direction (WD), vertical speed (VS), turbulence intensity (TI) and mixing layer height (MLH), retrieved by physical and proven algorithms. The PM concentration scales with the measured optical parameters and is also impacted by other environmental or meteorological parameters. Under the framework of the potential relationship between the PM concentration and the above parameters, the PM2.5 and PM10 concentration retrieval models were established using the stacking method of the ensemble learning technique; the models were trained using the in-situ data as true values, while the signal-to-noise ratio (SNR) at 1550 nm, δp,532, WS, WD, VS, the standard deviation of VS, TI, MLH provided by the Wind Flux 3000, as well as the relative humidity and temperature from ERA5, the hours of the day, and a "dust day" flag were used as inputs. The R2, RMSE, and MAE for the comparison between the predicted and true values of the PM2.5 test set are 0.857, 13.52 µg · m- 3, 9.16 µg · m- 3, and those of the PM10 test set are 0.926, 24.75 µg · m- 3, 14.39 µg · m- 3, respectively. The performance of the PM2.5 and PM10 concentration retrieval ensemble models is better than that of individual machine learning algorithms and better than that of the linear model. On 15th March 2023, a strong southeastward dust transport process with dust plume deposition was observed at Qingdao by the Wind Flux 3000. The analyses of the dust event demonstrated the Wind Flux 3000's ability to evaluate the transports of PM quantitatively.

Temporal Dynamics and Predictive Modelling of Streamflow and Water Quality Using Advanced Statistical and Ensemble Machine Learning Techniques

Temporal Dynamics and Predictive Modelling of Streamflow and Water Quality Using Advanced Statistical and Ensemble Machine Learning Techniques

Development of ensemble learning techniques and sequential model-based optimization for enhancing the generalizability of shale wettability predictions

Quantifying the wettability of shales is important for reservoir exploration and evaluation, as well as CO2 storage. Conventional experimental measurements are time-consuming and costly, while novel machine learning (ML) algorithms are challenged to predict shale wettability with accuracy and generalizability. Therefore, this study introduces an advanced shale wettability prediction framework by fusing ensemble learning algorithm and automatic hyperparameters optimization schemes (AHOS). Specifically, we firstly collected data from various gas geo-storage conditions from the literature. Then, an ensemble learning algorithm was utilized to extract complex nonlinear relationships among influential variables and shale wettability. During the training process, multiple optimization schemes, including random search (RS), Bayesian optimization (BO), and sequential model-based optimization (SMBO), were used and compared to automatically find the optimal hyperparameters of a wettability prediction model. The results show that AHOS can automatically and efficiently obtain the optimal hyperparameters, thus improves model's generalization. In addition, the prediction model based on AHOS process can accurately predict shale wettability. Through sensitivity analysis, we also revealed that the TOC and gas type are the top two most important influencing factors, which provides insights for CO2 sequestration. Moreover, we validated that the CatBoost model outperforms other tree-based models in predicting shale wettability and the SMBO method can provide a more effective result. In summary, the proposed shale wettability prediction framework is promising and efficient, which contributes to the exploration, evaluation, and development of shale oil reservoirs, further benefit to CO2 storage.

Using ensemble learning and hierarchical strategy to predict the outcomes of ESWL for upper ureteral stone treatment

Urinary tract stones are a common and frequently recurring medical issue. Accurately predicting the success rate after surgery can help avoid ineffective medical procedures and reduce unnecessary healthcare costs. This study collected data from patients with upper ureter stones who underwent extracorporeal shock wave lithotripsy, including cases of successful as well as unsuccessful stone removal after the first and second lithotripsy procedures, and constructed prediction systems for the outcomes of the first and second lithotripsy procedures. Features were extracted from three categories of information: patient characteristics, stone characteristics, and extracorporeal shock wave lithotripsy machine data, and additional features were created using Feature Creation. Finally, the impact of features on the models was analyzed using six methods to calculate feature importance. Our prediction model for the first lithotripsy, selected from among 43 methods and seven ensemble learning techniques, achieves an AUC of 0.91. For the second lithotripsy, the AUC reaches 0.76. The results indicate that the detailed and binary information provided by patients regarding their history of stone experiences contributes differently to the predictive accuracy of the first and second lithotripsy procedures. The prediction tool is available at https://predictor.isu.edu.tw/ks.

A stacking ensemble model for predicting the occurrence of carotid atherosclerosis.

Carotid atherosclerosis (CAS) is a significant risk factor for cardio-cerebrovascular events. The objective of this study is to employ stacking ensemble machine learning techniques to enhance the prediction of CAS occurrence, incorporating a wide range of predictors, including endocrine-related markers. Based on data from a routine health check-up cohort, five individual prediction models for CAS were established based on logistic regression (LR), random forest (RF), support vector machine (SVM), extreme gradient boosting (XGBoost) and gradient boosting decision tree (GBDT) methods. Then, a stacking ensemble algorithm was used to integrate the base models to improve the prediction ability and address overfitting problems. Finally, the SHAP value method was applied for an in-depth analysis of variable importance at both the overall and individual levels, with a focus on elucidating the impact of endocrine-related variables. A total of 441 of the 1669 subjects in the cohort were finally diagnosed with CAS. Seventeen variables were selected as predictors. The ensemble model outperformed the individual models, with AUCs of 0.893 in the testing set and 0.861 in the validation set. The ensemble model has the optimal accuracy, precision, recall and F1 score in the validation set, with considerable performance in the testing set. Carotid stenosis and age emerged as the most significant predictors, alongside notable contributions from endocrine-related factors. The ensemble model shows enhanced accuracy and generalizability in predicting CAS risk, underscoring its utility in identifying individuals at high risk. This approach integrates a comprehensive analysis of predictors, including endocrine markers, affirming the critical role of endocrine dysfunctions in CAS development. It represents a promising tool in identifying high-risk individuals for the prevention of CAS and cardio-cerebrovascular diseases.