Support Vector Machine Research Articles

Diagnosis of Sacroiliitis Through Semi-Supervised Segmentation and Radiomics Feature Analysis of MRI Images.

Sacroiliitis is a hallmark of ankylosing spondylitis (AS), and early detection plays an important role in managing the condition effectively. MRI is commonly used for diagnosing sacroiliitis, traditional methods often depend on subjective interpretation or limited automation which can introduce variability in diagnoses. The integration of semi-supervised segmentation and radiomics features may reduce reliance on expert interpretation and the need for large annotated datasets, potentially enhancing diagnostic workflows. To develop a diagnostic model for sacroiliitis and bone marrow edema (BME) using semi-supervised segmentation and radiomics analysis of MRI images. Retrospective cohort study. A total of 257 patients (161 males, 93 females; age 11-74 years), including 155 sacroiliitis and 175 BME patients. A total of 514 sacroiliac joint (SIJ) MRI images are analyzed, with 359 used for training and 155 for testing. 3.0 T, spin echo T1-weighted imaging (T1WI) and short-tau inversion recovery (STIR). SIJ segmentation is automated using the semi-supervised segmentation-based Unimatch framework. Manual delineation of SIJ regions of interest (ROIs) on T1WI images by an experienced radiologist (W.M., 10-year experience) served as the reference standard for segmentation performance evaluation. Radiomics features from T1WI and STIR are used to train machine learning models, including support vector machine (SVM), logistic regression (LR), and light gradient boosting machine (LightGBM), for sacroiliitis and BME detection. Performance is assessed using area under the curve (AUC), sensitivity, specificity, and accuracy. The Dice coefficient is used to assess the performance of the semi-supervised segmentation model on SIJ segmentation. Performance is evaluated using receiver operating characteristic (ROC) curves and decision curve analysis (DCA). The Unimatch model achieves an average Dice coefficient of 0.859 for SIJ segmentation. AUCs for sacroiliitis detection are 0.84 (LR), 0.86 (SVM), and 0.78 (LightGBM), while for BME detection, AUCs are 0.73 (LR), 0.76 (SVM), and 0.70 (LightGBM). This study demonstrates that semi-supervised segmentation combined with radiomics features and machine learning models provides a promising approach for diagnosis of sacroiliitis and BME. This study aimed to improve the diagnosis of sacroiliitis and bone marrow edema in patients with ankylosing spondylitis. The researchers used a method that automatically segments MRI images and analyzes features from those images. By applying machine learning, they created models to help detect sacroiliitis and bone marrow edema more accurately. The results show that this approach can effectively assist in identifying these conditions, with the best accuracy for sacroiliitis and bone marrow edema reaching 81.2% and 74.2%, respectively. This method could help doctors make better decisions, offering a promising tool for improving diagnosis in clinical settings. 3 TECHNICAL EFFICACY: Stage 2.

A comparison of different machine learning classifiers in predicting xerostomia and sticky saliva due to head and neck radiotherapy using a multi-objective, multimodal radiomics model.

Background and Purpose. Although radiotherapy techniques are a primary treatment for head and neck cancer (HNC), they are still associated with substantial toxicity and side effects. Machine learning (ML) based radiomics models for predicting toxicity mostly rely on features extracted from pre-treatment imaging data. This study aims to compare different models in predicting radiation-induced xerostomia and sticky saliva in both early and late stages HNC patients using CT and MRI image features along with demographics and dosimetric information.Materials and Methods.A cohort of 85 HNC patients who underwent radiation treatment was evaluated. We built different ML-based classifiers to build a multi-objective, multimodal radiomics model by extracting 346 different features from patient data. The models were trained and tested for prediction, utilizing Relief feature selection method and eight classifiers consisting eXtreme Gradient Boosting (XGBoost), Multilayer Perceptron (MLP), Support Vector Machines (SVM), Random Forest (RF), K-Nearest Neighbor (KNN), Naive Bayes (NB), Logistic Regression (LR), and Decision Tree (DT). The performance of the models was evaluated using sensitivity, specificity, area under the curve (AUC), and accuracy metrics.Results.Using a combination of demographics, dosimetric, and image features, the SVM model obtained the best performance with AUC of 0.77 and 0.81 for predicting early sticky saliva and xerostomia, respectively. Also, SVM and MLP classifiers achieved a noteworthy AUC of 0.85 and 0.64 for predicting late sticky saliva and xerostomia, respectively.Conclusion. This study highlights the potential of baseline CT and MRI image features, combined with dosimetric data and patient demographics, to predict radiation-induced xerostomia and sticky saliva. The use of ML techniques provides valuable insights for personalized treatment planning to mitigate toxicity effects during radiation therapy for HNC patients.

Tree Species Classification Based on Point Cloud Completion

LiDAR is an active remote sensing technology widely used in forestry applications, such as forest resource surveys, tree information collection, and ecosystem monitoring. However, due to the resolution limitations of 3D-laser scanners and the canopy occlusion in forest environments, the tree point clouds obtained often have missing data. This can reduce the accuracy of individual tree segmentation, which subsequently affects the tree species classification. To address the issue, this study used point cloud data with RGB information collected by the UAV platform to improve tree species classification by completing the missing point clouds. Furthermore, the study also explored the effects of point cloud completion, feature selection, and classification methods on the results. Specifically, both a traditional geometric method and a deep learning-based method were used for point cloud completion, and their performance was compared. For the classification of tree species, five machine learning algorithms—Random Forest (RF), Support Vector Machine (SVM), Back Propagation Neural Network (BPNN), Quadratic Discriminant Analysis (QDA), and K-Nearest Neighbors (KNN)—were utilized. This study also ranked the importance of features to assess the impact of different algorithms and features on classification accuracy. The results showed that the deep learning-based completion method provided the best performance (avgCD = 6.14; avgF1 = 0.85), generating more complete point clouds than the traditional method. On the other hand, compared with SVM and BPNN, RF showed better performance in dealing with multi-classification tasks with limited training samples (OA-87.41%, Kappa-0.85). Among the six dominant tree species, Pinus koraiensis had the highest classification accuracy (93.75%), while that of Juglans mandshurica was the lowest (82.05%). In addition, the vegetation index and the tree structure parameter accounted for 50% and 30%, respectively, in the top 10 features in terms of feature importance. The point cloud intensity also had a high contribution to the classification results, indicating that the lidar point cloud data can also be used as an important basis for tree species classification.

Investigation of Engine Lubrication Oil Quality Using a Support Vector Machine and Electronic Nose

Monitoring the quality of engine oil improves engine efficiency and reduces engine maintenance costs. Several methods have been proposed for this purpose; however, most of them take too long to test oil quality. This paper introduces a fast, simple, and accurate method to determine oil quality using an electronic nose and artificial intelligence. The TU5 engine and 10-40W “Behran Super Pishtaz” engine oil were used in the experiments. Tests were conducted at six different quality levels. Oil properties such as viscosity, density, flash point, and freezing point were measured at each level. Additionally, oil smell signals were recorded using an olfactory machine at these quality levels. The fraction method was employed to adjust the sensors’ responses. Five statistical features were extracted from each signal, and these features were used to train and test a support vector machine (SVM) for classifying oil quality using the five-fold cross-validation method. The results indicated a statistically significant change in viscosity and density with variations in oil quality. The density increased as the quality decreased. Viscosity, however, initially decreased and then increased at later stages. An analysis of the sensory outputs revealed that changes in oil quality also affected these outputs, with the most pronounced sensitivity observed in the MQ135 and MQ138 sensors. The final accuracies of the SVM in classifying oil quality were 68.22%, 85.86%, and 95.44% for linear, radial basis function (RBF), and polynomial kernels, respectively. The SVM sensitivities for oil qualities A, B, C, D, E, and F were 97.99%, 97.37%, 95.51%, 92.67%, 94.48%, and 94.59%, respectively.

Rice yield predictions using remote sensing and machine learning algorithms: A review

Crop yield prediction is becoming increasingly crucial due to global food security concerns, as highlighted by recent reports from the World Health Organization. Accurate early predictions can mitigate famine risks by estimating food supply, which is essential for 820 million people facing hunger globally. Rice is the primary staple food consumed worldwide; therefore, global rice yield and rice area are monitored using emerging technologies such as remote sensing (RS) and machine learning (ML). These technologies provide valuable tools for enhancing rice yield predictions. RS includes critical information on crop health, soil conditions and weather patterns. In contrast, ML algorithms analyze these datasets to identify patterns and relationships that affect yield. Integrating these technologies offers promising improvements in yield forecasting accuracy, with applications showing successful yield predictions 1-3 months before harvest. Various ML techniques, including Random Forest, Support Vector Machines and deep learning models such as LSTM (Long-Short Term Memory), have been employed, often in combination with RS data. However, these models face challenges, such as data quality, managing high-dimensional RS data and accounting for spatial and temporal variability. Despite these challenges, integrating RS and ML has significant potential for advancing precision agriculture and achieving sustainable food production. This study explores the advancements, practical applications and challenges associated with using RS and ML for rice yield prediction, emphasizing the importance of these technologies in addressing global food security and promoting sustainable agricultural practices.

Geospatial digital mapping of soil organic carbon using machine learning and geostatistical methods in different land uses

Improper management of soil resources leads to the destruction of soil organic carbon (SOC) stock and, as a result, the reduction of soil quality, as well as accelerating the process of climate change through the release of SOC into the atmosphere. This study was conducted to evaluate potential of different simulation models to map the spatial variability of SOC as affected by land use in a study area in Qarasu watershed in Kermanshah province, west of Iran. Map of sampling points was prepared using the Latin hypercube sampling method. A total of 168 observation points were selected and the soil profile was dug and described in these points. The soil samples were taken in different horizon to determine the SOC content in laboratory. SOC was mapped using kriging geostatistical method in the study area. The SOC changes were simulated using multivariate analysis and machine learning methods including generalized linear model (GLM), linear additive model (LAM), cubist, random forest (RF), and support vector machine (SVM) models. Comprehensive measurement data is utilized to develop and validate the predictive models. Predictor variables included 16 topographic variables as well as two vegetation, six parent material, and four climatic variables. In-depth statistical analyses are proposed to evaluate the proposed models performance. The results showed that the SOC content ranged from 0.19 to 8.44 percent in different land uses. The spherical variogram model with MAE = 0.41 best fits to interpolate and map SOC using ordinary kriging in different land uses. The LAM method was estimated wider range of the SOC change (SOC = 0.18–4.82%) among different simulation model. However, the RF model (R2 = 0.64 and RMSE = 0.58%) was the most accurate in predicting SOC quantity comparing the other simulation models. It can be used as a reliable simulation model to predict SOC variability in the study area and other similar semiarid regions of West Asia in different land uses. Among the different predictor variables, the parent material’s intrinsic soil properties and topography variables had the greatest effect in predicting soil organic carbon variability.

LeafLaminaMap: Exploring Leaf Color Patterns Using RGB Color Indices

The color of the plant leaves is a major concern in many areas of agriculture. Pigmentation and its pattern provide the possibility to distinguish genotypes and a basis for annual crop management practices. For example, the nutrient and water status of plants is reflected in the chlorophyll content of leaves that are strongly linked to the lamina coloration. Pests and diseases (virus or bacterial infections) also cause symptoms on the foliage. These symptoms induced by biotic and abiotic stressors often have a specific pattern, which allows for their prediction based on remote sensing. In this report, an RGB (red, green and blue) image processing system is presented to determine leaf lamina color variability based on RGB-based color indices. LeafLaminaMap was developed in Scilab with the Image Processing and Computer Vision toolbox, and the code is available freely at GitHub. The software uses RGB images to visualize 29 color indices and the R, G and B values on the lamina, as well as to calculate the statistical parameters. In this case study, symptomatic (senescence, fungal infection, etc.) and healthy grapevine (Vitis vinifera L.) leaves were collected, digitalized and analyzed with the LeafLaminaMap software according to the mean, standard deviation, contrast, energy and entropy of each channel (R, G and B) and color index. As an output for each original image in the sample set, the program generates 32 images, where each pixel is constructed using index values calculated from the RGB values of the corresponding pixel in the original image. These generated images can subsequently be used to help the end-user identify locally occurring symptoms that may not be visible in the original RGB image. The statistical evaluation of the samples showed significant differences in the color pattern between the healthy and symptomatic samples. According to the F value of the ANOVA analysis, energy and entropy had the largest difference between the healthy and symptomatic samples. Linear discriminant analysis (LDA) and support vector machine (SVM) analysis provided a perfect recognition in calibration and confirmed that energy and entropy have the strongest discriminative power between the healthy and symptomatic samples. The case study showed that the LeafLaminaMap software is an effective environment for the leaf lamina color pattern analysis; moreover, the results underline that energy and entropy are valuable features and could be more effective than the mean and standard deviation of the color properties.

Advancing Optical Coherence Tomography Diagnostic Capabilities: Machine Learning Approaches to Detect Autoimmune Inflammatory Diseases.

In many parts of the world including India, the prevalence of autoimmune inflammatory diseases such as neuromyelitis optica spectrum disorders (NMOSD), myelin oligodendrocyte glycoprotein antibody-associated disease (MOGAD), and multiple sclerosis (MS) is rising. A diagnosis is often delayed due to insufficient diagnostic tools. Machine learning (ML) models have accurately differentiated eyes of patients with MS from those of healthy controls (HCs) using optical coherence tomography (OCT)-based retinal images. Examining OCT characteristics may allow for early differentiation of these conditions. The objective of this study was to determine feasibility of ML analyses to distinguish between patients with different autoimmune inflammatory diseases, other ocular diseases, and HCs based on OCT measurements of the peripapillary retinal nerve fiber layer (pRNFL), ganglion cell-inner plexiform layer (GCIPL), and inner nuclear layers (INLs). Eyes of people with MS (n = 99 patients), NMOSD (n = 40), MOGAD (n = 74), other ocular diseases (OTHER, n = 16), and HCs (n = 54) from the Mangalore Demyelinating Disease Registry were included. Support vector machine (SVM) classification models incorporating age, pRNFL, GCIPL, and INL were performed. Data were split into training (70%) and testing (30%) data and accounted for within-patient correlations. Cross-validation was used in training to choose the best parameters for the SVM model. Accuracy and area under receiver operating characteristic curves (AUROCs) were used to assess model performance. The SVM models distinguished between eyes of patients with each condition (i.e., MOGAD vs NMOSD, NMOSD vs HC, MS vs OTHER, etc) with strong discriminatory power demonstrated from the AUROCs for these comparisons ranging from 0.81 to 1.00. These models also performed with moderate to high accuracy, ranging from 0.66 to 0.81, with the exception of the MS vs NMOSD comparison, which had an accuracy of 0.53. ML models are useful for distinguishing between autoimmune inflammatory diseases and for distinguishing these from HCs and other ocular diseases based on OCT measures. This study lays the groundwork for future deep learning studies that use analyses of raw OCT images for identifying eyes of patients with such disorders and other etiologies of optic neuropathy.

Adversarial Attacks on Supervised Energy-Based Anomaly Detection in Clean Water Systems

Critical National Infrastructure includes large networks such as telecommunications, transportation, health services, police, nuclear power plants, and utilities like clean water, gas, and electricity. The protection of these infrastructures is crucial, as nations depend on their operation and stability. However, cyberattacks on such systems appear to be increasing in both frequency and severity. Various machine learning approaches have been employed for anomaly detection in Critical National Infrastructure, given their success in identifying both known and unknown attacks with high accuracy. Nevertheless, these systems are vulnerable to adversarial attacks. Hackers can manipulate the system and deceive the models, causing them to misclassify malicious events as benign, and vice versa. This paper evaluates the robustness of traditional machine learning techniques, such as Support Vector Machines (SVMs) and Logistic Regression (LR), as well as Artificial Neural Network (ANN) algorithms against adversarial attacks, using a novel dataset captured from a model of a clean water treatment system. Our methodology includes four attack categories: random label flipping, targeted label flipping, the Fast Gradient Sign Method (FGSM), and Jacobian-based Saliency Map Attack (JSMA). Our results show that, while some machine learning algorithms are more robust to adversarial attacks than others, a hacker can manipulate the dataset using these attack categories to disturb the machine learning-based anomaly detection system, allowing the attack to evade detection.

Research on Rapid and Non-Destructive Detection of Coffee Powder Adulteration Based on Portable Near-Infrared Spectroscopy Technology

This study explores the feasibility of using portable near-infrared spectroscopy for the rapid and non-destructive detection of coffee adulteration. Spectral data from adulterated coffee samples in the 900–1700 nm range were collected and processed using five preprocessing methods. For qualitative detection, the Support Vector Machine (SVM) algorithm was applied. For quantitative detection, two optimization algorithms, Invasive Weed Optimization (IWO) and Binary Chimp Optimization Algorithm (BChOA), were used for the feature wavelength selection. The results showed that convolution smoothing combined with multiple scattering correction effectively improved the signal-to-noise ratio. SVM achieved 96.88% accuracy for qualitative detection. For the quantitative analysis, the IWO algorithm identified key wavelengths, reducing data dimensionality by 82.46% and improving accuracy by 10.96%, reaching 92.25% accuracy. In conclusion, portable near-infrared spectroscopy technology can be used for the rapid and non-destructive qualitative and quantitative detection of coffee adulteration and can serve as a foundation for the further development of rapid, non-destructive testing devices. At the same time, this method has broad application potential and can be extended to various food products such as dairy, juice, grains, and meat for quality control, traceability, and adulteration detection. Through the feature wavelength selection method, it can effectively identify and extract spectral features associated with these food components (such as fat, protein, or characteristic compounds), thereby improving the accuracy and efficiency of detection, further ensuring food safety and enhancing the level of food quality control.

The relationship between activities of daily living and speech impediments based on evidence from statistical and machine learning analyses

IntroductionSpeech impediments (SIs) are increasingly prevalent among middle-aged and older adults, raising concerns within public health. Early detection of potential SI in this demographic is critical. This study investigates the potential of Activities of Daily Living (ADL) as a predictive marker for SI, utilizing data from the 2018 China Health and Retirement Longitudinal Study (CHARLS), which includes 10,136 individuals aged 45 and above. The Barthel Index (BI) was used to assess ADL, and the correlation between ADL and SI was examined through statistical analyses. Machine learning algorithms (Support Vector Machine, Decision Tree, and Logistic Regression) were employed to validate the findings and elucidate the underlying relationship between ADL and SI.BackgroundSI poses significant challenges to the health and quality of life of middle-aged and older adults, increasing the demands on community-based and home care services. In the context of global aging, it is crucial to investigate the factors contributing to SI. While the role of ADL as a potential biomarker for SI remains unclear, this study aims to provide new evidence supporting ADL as an early predictor of SI through statistical analysis and machine learning validation.MethodsData were derived from the 2018 CHARLS national baseline survey, comprising 10,136 participants aged 45 and above. ADL was evaluated using the BI, and SI was assessed based on the CHARLS records of “Speech impediments.” Statistical analyses, including independent sample t-tests, chi-square tests, Pearson and Spearman correlation tests, and hierarchical multiple linear regression, were conducted using SPSS 25.0. Machine learning algorithms, specifically Support Vector Machine (SVM), Decision Tree (DT), and Logistic Regression (LR), were implemented in Python 3.10.2.ResultsAnalysis of demographic characteristics revealed that the average BI score in the “With Speech impediments” group was 49.46, significantly lower than the average score of 85.11 in the “Without Speech impediments” group. Pearson correlation analysis indicated a significant negative correlation between ADL and SI (r = −0.205, p < 0.001). Hierarchical multiple linear regression confirmed the robustness of this negative correlation across three models (B = −0.001, β = −0.168, t = −16.16, 95% CI = −0.001 to −0.001, p = 0.000). Machine learning algorithms validated the statistical findings, confirming the predictive accuracy of ADL for SI, with the area under the curve (AUC) scores of SVM-AUC = 0.648, DT-AUC = 0.931, and LR-AUC = 0.666. The inclusion of BI in the models improved the overall predictive performance, highlighting its positive impact on SI prediction.ConclusionThe study employed various statistical methodologies to demonstrate a significant negative correlation between ADL and SI, a finding further corroborated by machine learning algorithms. Impairment in ADL increases the likelihood of SI occurrence, underscoring the importance of maintaining ADL in middle-aged and older populations to mitigate the risk of SI.

Machine Learning and Deep Learning Models for Dengue Diagnosis Prediction: A Systematic Review

The global crisis triggered by the dengue outbreak has increased mortality and placed significant pressure on healthcare services worldwide. In response to this crisis, there has been a notable increase in research employing machine learning and deep learning algorithms to anticipate diagnosis in patients with suspected dengue. To conduct a comprehensive systematic review, a detailed analysis was carried out to explore and examine the machine learning methodologies applied in diagnosing this disease. An exhaustive search was conducted across numerous scientific databases, including Scopus, IEEE Xplore, PubMed, ACM, ScienceDirect, Wiley, and Sage, encompassing studies up to May 2024. This extensive search yielded a total of 2723 relevant articles. Following a rigorous evaluation, 32 scientific studies were selected for the final review, meeting the established criteria. A comprehensive analysis of these studies revealed the implementation of 48 distinct machine learning and deep learning algorithms, showcasing the heterogeneity of methodological approaches employed in the research domain. The results indicated that, in terms of performance, the support vector machine (SVM) algorithm was the most efficient, being reported in 25% of the analyzed studies. The Random Forest algorithm was the second most frequently used, appearing in 15.62% of the 32 reviewed articles. The PCA-SVM algorithm (poly-5), a variant of SVM, emerged as the best-performing model, achieving 99.52% accuracy, 99.75% sensitivity, and 99.09% specificity. These findings offer significant insights into the potential of machine learning techniques in the early diagnosis of dengue, underscoring the necessity to persist in exploring and refining these methodologies to enhance clinical care in cases of this disease.

A fine-tuned adaptive weight deep dense meta stacked transfer learning model for effective cervical cancer prediction

Abstract In the digital world of remarkable technological advancements, the detection of cervical cancer at early stages is of important clinical significance as it can vastly improve the survival rate during treatment. Cervical cytopathology, often known as a Pap test is the frequently adopted screening method for cervical cancer. However, the test seems to be effective but investigation of images containing Pap smear with the help of a microscope is a difficult as well as laborious exercise. The procedure for the same demands an expert in the area and is often time-consuming. The serious pitfalls in subjective clinical evaluation evoke the need of developing an automated system for more reliable cervical cancer diagnosis. Therefore, the goal of this study primarily focuses on designing a Deep learning model to process the Pap smear images and correctly classify the cervical cells. For this purpose, firstly, a publically available dataset namely SIPaKMeD is utilized. Then, different data pre-processing methods are applied to intensify the data quality for effective analysis. Next, a novel stacking model is proposed that leverages a Support Vector Classifier (SVC) as a Meta model over a combination of different Transfer Learning Models including VGG16, ResNet101, InceptionV3, Xception, DenseNet169, and Inception ResNet. Furthermore, the dense layers are added to tune the underlying base transfer learning models to learn fine-tuned adaptive weights. The results obtained from experimental evaluation demonstrate the efficacy of the proposed stacking model by yielding the highest accuracy rate of 95.66% in comparison to other employed methods and existing state-of-the-art techniques.

Research on the Circuit Breaker State Judgment Method Based on Ant Colony Optimization Dempster-Shafer Evidence Theory

The paper proposes a method for circuit breaker state judgment based on ant colony algorithm-optimized Dempster-Shafer evidence theory.It can improve the accuracy and robustness of state judgment. As a key device in the power system, the state judgment of circuit breakers is crucial for the safety and stability of the power grid. Existing methods have limitations in handling conflicts and uncertainties of multi-source data, and a single model is difficult to meet the needs of complex data fusion. Therefore, the paper applies the ant colony algorithm to optimize the basic probability assignment in Dempster-Shafer evidence theory to improve the fusion effect of multi-source data. The ant colony algorithm, through its global search and adaptive characteristics, can effectively optimize evidence combination and enhance the accuracy of the fusion results. The experiment used a support vector machine model based on current signals and a decision tree model based on vibration signals for data fusion and discrimination. The results showed that the Dempster-Shafer evidence theory model optimized by the ant colony algorithm achieved a discrimination accuracy of 75% under various circuit breaker conditions. Compared to the Dempster-Shafer evidence theory fusion model, it improved by approximately 8.3%, and compared to the current research’s Dempster-Shafer evidence theory and neural network methods, it improved by 5%.. This method has broad application prospects in enhancing the operational stability of power grid equipment and improving fault detection efficiency.

A logistic regression model to predict long-term survival for borderline resectable pancreatic cancer patients with upfront surgery

BackgroundThe machine learning model, which has been widely applied in prognosis assessment, can comprehensively evaluate patient status for accurate prognosis classification. There still has been a debate about which predictive strategy is better in patients with borderline resectable pancreatic cancer (BRPC). In the present study, we establish a logistic regression model, aiming to predict long-term survival and identify related prognostic factors in patients with BRPC who underwent upfront surgery.MethodsMedical records of patients with BRPC who underwent upfront surgery with portal vein resection and reconstruction from Jan. 2011 to Dec. 2020 were reviewed. Based on postoperative overall survival (OS), patients were divided into the short-term group (≤ 2 years) and the long-term group (> 2 years). Univariate and multivariate analyses were performed to compare perioperative variables and long-term prognoses between groups to identify related independent prognostic factors. All patients are randomly divided into the training set and the validation set at a 7:3 ratio. The logistic regression model was established and evaluated for accuracy through the above variables in the training set and the validation set, respectively, and was visualized by Nomograms. Meanwhile, the model was further verified and compared for accuracy, the area under the curve (AUC) of the receiver operating characteristic curves (ROC), and calibration analysis. Then, we plotted and sorted perioperative variables by SHAP value to identify the most important variables. The first 4 most important variables were compared with the above independent prognostic factors. Finally, other models including support vector machines (SVM), random forest, decision tree, and XGBoost were also constructed using the above 4 variables. 10-fold stratified cross-validation and the AUC of ROC were performed to compare accuracy between models.Results104 patients were enrolled in the study, and the median OS was 15.5 months, the 0.5-, 1-, and 2- years OS were 81.7%, 57.7%, and 30.8%, respectively. In the long-term group (n = 32) and short-term group (n = 72), the overall median survival time and the 1-, 2-, 3- years overall survival were 38 months, 100%, 100%, 61.3% and 10 months, 38.9%, 0%, 0%, respectively. 4 variables, including age, vascular invasion length, vascular morphological malformation, and local lymphadenopathy were confirmed as independent risk factors between the two groups following univariate and multivariate analysis. The AUC between the training set (n = 72) and the validation set (n = 32) were 0.881 and 0.875. SHAP value showed that the above variables were the first 4 most important. The AUC following 10-fold stratified cross-validation in the logistic regression (0.864) is better than SVM (0.693), random forest (0.789), decision tree (0.790), and XGBoost (0.726).ConclusionAge, vascular invasion length, vascular morphological malformation, and local lymphadenopathy were independent risk factors for long-term survival of BRPC patients with upfront surgery. The logistic regression model plays a predictive role in long-term survival and may further assist surgeons in deciding the treatment option for BRPC patients.

Modeling of Forest Fire Risk Areas of Amazonas Department, Peru: Comparative Evaluation of Three Machine Learning Methods

Forest fires are the result of poor land management and climate change. Depending on the type of the affected eco-system, they can cause significant biodiversity losses. This study was conducted in the Amazonas department in Peru. Binary data obtained from the MODIS satellite on the occurrence of fires between 2010 and 2022 were used to build the risk models. To avoid multicollinearity, 12 variables that trigger fires were selected (Pearson ≤ 0.90) and grouped into four factors: (i) topographic, (ii) social, (iii) climatic, and (iv) biological. The program Rstudio and three types of machine learning were applied: MaxENT, Support Vector Machine (SVM), and Random Forest (RF). The results show that the RF model has the highest accuracy (AUC = 0.91), followed by MaxENT (AUC = 0.87) and SVM (AUC = 0.84). In the fire risk map elaborated with the RF model, 38.8% of the Amazonas region possesses a very low risk of fire occurrence, and 21.8% represents very high-risk level zones. This research will allow decision-makers to improve forest management in the Amazon region and to prioritize prospective management strategies such as the installation of water reservoirs in areas with a very high-risk level zone. In addition, it can support awareness-raising actions among inhabitants in the areas at greatest risk so that they will be prepared to mitigate and control risk and generate solutions in the event of forest fires occurring under different scenarios.

Developing a Dataset of Audio Features to Classify Emotions in Speech

Emotion recognition in speech has gained increasing relevance in recent years, enabling more personalized interactions between users and automated systems. This paper presents the development of a dataset of features obtained from RAVDESS (Ryerson Audio-Visual Database of Emotional Speech and Song) to classify emotions in speech. The paper highlights audio processing techniques such as silence removal and framing to extract features from the recordings. The features are extracted from the audio signals using spectral techniques, time-domain analysis, and the discrete wavelet transform. The resulting dataset is used to train a neural network and the support vector machine learning algorithm. Cross-validation is employed for model training. The developed models were optimized using a software package that performs hyperparameter tuning to improve results. Finally, the emotional classification outcomes were compared. The results showed an emotion classification accuracy of 0.654 for the perceptron neural network and 0.724 for the support vector machine algorithm, demonstrating satisfactory performance in emotion classification.

Development and External Validation of [18F]FDG PET-CT-Derived Radiomic Models for Prediction of Abdominal Aortic Aneurysm Growth Rate

Objective (1): To develop and validate a machine learning (ML) model using radiomic features (RFs) extracted from [18F]FDG PET-CT to predict abdominal aortic aneurysm (AAA) growth rate. Methods (2): This retrospective study included 98 internal and 55 external AAA patients undergoing [18F]FDG PET-CT. RFs were extracted from manual segmentations of AAAs using PyRadiomics. Recursive feature elimination (RFE) reduced features for model optimisation. A multi-layer perceptron (MLP) was developed for AAA growth prediction and compared against Random Forest (RF), XGBoost, and Support Vector Machine (SVM). Accuracy was evaluated via cross-validation, with uncertainty quantified using dropout (MLP), standard deviation (RF), and 95% prediction intervals (XGBoost). External validation used independent data from two centres. Ground truth growth rates were calculated from serial ultrasound (US) measurements or CT volumes. Results (3): From 93 initial RFs, 29 remained after RFE. The MLP model achieved an MAE ± SEM of 1.35 ± 3.2e−4 mm/year with the full feature set and 1.35 ± 2.5e−4 mm/year with RFE. External validation yielded 1.8 ± 8.9e−8 mm/year. RF, XGBoost, and SVM models produced comparable accuracies internally (1.4–1.5 mm/year) but showed higher errors during external validation (1.9–1.97 mm/year). The MLP model demonstrated reduced uncertainty with the full feature set across all datasets. Conclusions (4): An MLP model leveraging [18F]FDG PET-CT radiomics accurately predicted AAA growth rates and generalised well to external data. In the future, more sophisticated stratification could guide individualised patient care, facilitating risk-tailored management of AAAs.

Quantum adversarial learning for kernel methods

We show that hybrid quantum classifiers based on quantum kernel methods and support vector machines are vulnerable against adversarial attacks, namely small engineered perturbations of the input data can deceive the classifier into predicting the wrong result. Nonetheless, we also show that simple defense strategies based on data augmentation with a few crafted perturbations can make the classifier robust against new attacks. Our results find applications in security-critical learning problems and in mitigating the effect of some forms of quantum noise, since the attacker can also be understood as part of the surrounding environment.

Nondestructive TVB‐N Detection in Packaged Beef Using Hyperspectral Imaging

ABSTRACTHyperspectral imaging (HSI) technology has shown potential in monitoring fresh meat. It is necessary to investigate the rapid detection technology of packaged fresh meat in order to ensure food quality and safety. In this study, hyperspectral images of beef in the presence of packaging film were collected, and the spectra were optimized using spectral pretreatment, band selection and film radiation correction methods to predict the total volatile basic nitrogen (TVB‐N) of beef in the presence of packaging film. The results of the competitive adaptive reweighted sampling (CARS) combined with least squares support vector machines (CARS‐LSSVM) model constructed by using the corrected reflectance spectra were = 0.8687, RMSEP = 3.1970 mg/100 g, which is higher than that obtained before the correction ( was 0.8168, and RMSEP was 3.0449 mg/100 g). The results reveal that HSI with radiation correction has the potential in monitoring the TVB‐N content of beef through packaging film.