An exact penalty method with nonmonotone line search and rapid infeasibility detection for constrained multiobjective optimization: Application in supervised machine learning

Subcutaneous tissue structural feature identification using unsupervised machine learning.

Serum metabolomic signatures of relapse recovery in early multiple sclerosis.

Supervised machine learning methods for short-term prediction of a sudden cardiac death from electrocardiogram

Previous research suggests that adolescents with BPD (aBPD) exhibit distinct neuroanatomical alterations, although methodological limitations such as low sample size, and the reliance on univariate massive statistical analyses, prevent conclusions. Moreover, the possibility to associate these abnormalities with clinical features has been only partially explored. This study aims to investigate structural brain differences in the largest sample of adolescents with BPD to date, using a combination of unsupervised and supervised machine learning approaches: Source-Based Morphometry (SBM) and a deep neural network whose architecture was optimized through genetic algorithms. We hypothesize that adolescents with BPD will exhibit increased gray matter volume in the default mode network (DMN) and cerebellum, strictly related with emotion dysregulation and borderline symptoms, alongside reduced gray matter in frontal control networks. High-resolution T1-weighted structural MRI of 129 adolescents with BPD (aged 12-17) and 107 age-, gender-, and education-matched healthy controls (HCs) were analyzed using SBM to identify networks of covarying gray matter concentration (GMC). Borderline symptomatology, difficulties in emotion regulation, anxiety-related problems, and global functioning were assessed to characterize the meaning of neural findings. Compared to HCs, adolescents with BPD exhibited significantly increased GMC in regions overlapping with the posterior hub of the DMN, and the cerebellum, and reduced GMC in frontal control regions. Importantly, the GMC alterations inside the cerebellum and the DMN positively correlated with the difficulties in emotion regulation such as emotional clarity and emotion regulation difficulties, self-harm injuries, anxiety and depressive symptoms, and negatively correlated with global assessment functioning. The deep learning model confirmed these findings and provided a good generalization performance. Our findings suggest that gray matter alterations in regions ascribed to the default mode network, cerebellum, and frontal control regions play a crucial role in emotional regulation deficits and self-injurious behaviors in adolescent BPD. This study provides new insights into the neurobiological mechanisms of BPD in youth and offering potential biomarker and targets for treatments.

Exploring temperamental and clinical predictors of lithium treatment outcomes in bipolar disorder using diverse machine learning approaches.

The subject of Intrusion Detection System (IDS) is a very interesting research topic actively pursued by many investigators. The goal of intrusion detection is to monitor network assets and to detect anomalous behaviour and misuse. Intrusion Detection Systems aimto identify attacks with a high detection rate and a low false alarm rate. Intrusion Detection Systems (IDS) can monitor users, applications, networks, or combinations of the three, in order to detect well-known and unknown attacks. In this research work, I read many papers inwhich I found that some papers used supervised machine learning method. In this method, SVM algorithm was used and kernel function was also used. Using this concept, when intrusion detection system was built, its accuracy was between 70% to 81%. Apart from this, whenI run generic algorithm and another model based on signature, its performance was between 85% to 91%. Similarly, when I used generic algorithm and another model based on anomaly, its performance was the best. Apart from this, I run some other models in which unsupervised method was used. In this, FCM algorithm was used which follows DBSCAN algorithm.it reduce the false positive rate. In my research paper, I have used a hybrid method in which I have created an intrusion detection model using SVM, FCM and DBSCAN algorithm which not only reduces the false positive rate but also improves network security. Keywords: Data Mining, Intrusion Detection, Classification, False Positive, Anomaly based algorithm, Machine Learning, Deep Learning, NSL-KDD Data set.

The increasing demand for personalized healthcare solutions highlights the limitations of one-size-fits-all treatment strategies. Digital twin (DT) technology, which enables real-time virtual replicas of physical systems, offers a promising approach to advance personalized medicine through continuous monitoring, simulation, and prediction. This study presents the foundational phase of a machine-learning-driven biosensor DT designed to support personalized infertility treatment through integration with a Smart Health Monitoring System. The DT replicates the behavior of a field-effect transistor (FET)-based biosensor functionalized with 17$\beta$-estradiol aptamers and trained on experimental data obtained from silicon nanonet BioFET prototypes. Seven supervised machine learning algorithms were evaluated to predict hormone concentration from electrical parameters ($V_{g}$, $I_{sd}$). The K-Nearest Neighbors (KNN) model achieved the highest predictive accuracy ($R^{2} = 0.99$, $\text{CV}\text{-}R^{2} = 0.98$, RMSE = 11.87 pg/mL) and demonstrated robust cross-device generalization under Leave-One-Biosensor-Out validation ($R^{2} = 0.59$). These results confirm the model's capability to capture nonlinear relationships and generalize across independently fabricated sensors. The developed model constitutes a validated predictive core of a biosensor digital twin. At the current stage, the DT is limited to predictive modeling and does not yet implement real-time synchronization or closed-loop feedback, which are planned in future work. This study establishes a practical framework for data-driven digital twins of biosensors and demonstrates their potential for integration into smart health monitoring systems supporting personalized infertility care. The proposed approach provides a foundation for real-time, adaptive, and clinically relevant biosensor twins in precision medicine.

This study presents a structured multi-sensor predictive maintenance framework for CO2 laser cutting machines based on real-time data acquisition and supervised machine learning. The proposed architecture integrates heterogeneous sensor signals—including vibration, temperature, humidity, and acoustic measurements—through synchronized feature-level fusion to characterize machine operational states. A statistically grounded thresholding strategy, validated using two years of operational observations and controlled experimental perturbations, is employed to distinguish normal and abnormal behavior. Sensor data are processed using a Decision Tree classifier implemented in Python with Scikit-learn, enabling short-horizon probabilistic fault prediction during operational cycles. The system is deployed in a real industrial environment and validated using cross-validation and structured dataset partitioning to assess generalization performance. Results demonstrate reliable fault discrimination capability under controlled operational conditions, highlighting the effectiveness of feature-level sensor integration for early anomaly detection. The modular hardware–software architecture supports adaptability to other CNC platforms with appropriate recalibration and retraining. The proposed framework provides a low-cost, interpretable, and computationally efficient solution for real-time industrial predictive maintenance applications.

Power transformers are strategic and valuable assets in electrical systems, as their unexpected failures can lead to significant operational and financial losses for power sector companies and consumers. Although there have been advancements in monitoring their operational conditions, some methodologies still require specialized interpretation, lack standardization, or adopt models whose complexity can hinder integration with the usual operational practices of maintenance professionals. In this context, the objective of this work is to develop binary classifiers based on machine learning algorithms for fast and efficient prediction of the operational state of power transformers, labeled as Satisfactory or Unsatisfactory, using data derived from physicochemical tests, Dissolved Gas Analysis (DGA), and performance indices, based on real equipment samples. The methodology involves the development of supervised machine learning models, such as Random Forest, HistGradientBoosting, Balanced Logistic Regression, and XGBoost, implemented with stratified cross-validation. The results indicate that the classifiers can satisfactorily identify transformers in critical condition, even in a scenario with considerable data dispersion. Therefore, the proposed approach represents a promising tool for technical decision-making in preventive maintenance strategies, combining reliability, scalability, and ease of application in field environments.

Machine Learning to Characterize Speech Recognition and Quality of Life Outcomes in Adult Cochlear Implant Users.

About 21million teenagers became pregnant annually throughout the globe. Teen pregnancy is a serious issue in Sub-Saharan Africa, with East Africa reporting the highest rates. In the field of public health, machine learning has become an invaluable tool due to its ability to process large, complex datasets and identify trends. This study uses machine learning to predict and identify key determinants of teenage pregnancy in East Africa, utilizing DHS. A supervised machine learning approach, specifically the Random Forest algorithm, was applied to analyze relationships between predictors and teenage pregnancy outcomes. Data preprocessing included handling missing values, feature scaling, and addressing class imbalance using Tomek Links and SMOTE Model performance was evaluated using metrics such as accuracy, confusion matrix, and ROC AUC. The final model was validated on a separate test set to ensure generalizability and predictive accuracy. Random Forest demonstrated superior performance, with an AUC of 94.6, an accuracy of 89.1%, an F1 score of 89%, a recall of 88%, and a precision of 90%. Kenya had the highest rate of teenage pregnancies at 19.1%, with a 95% confidence interval of [18.12%, 20.08%]. Key predictors of teenage pregnancy in East Africa include maternal education, marital status, age at first sexual intercourse, wealth status, place of residence, distance to health facilities, and social media usage. These findings suggest that expanding reproductive health services in rural areas, with strengthened youth-friendly services; promoting education about teenage pregnancy through social media; and integrating reproductive health education into school curricula may decrease teenage pregnancy in East Africa.

Electric cooperatives in the Philippines remain vital to rural electrification, yet many continue to rely on manual complaint-handling methods such as walk-in transactions, phone calls, and handwritten logbooks. These practices often result in delayed responses, fragmented documentation, and weak accountability, which in turn erode consumer trust. This study designed, developed, and evaluated a Consumer Complaint Management System (CCMS) enhanced with artificial intelligence (AI)-based segmentation to address these challenges. The system was built using the Agile Development Model, allowing iterative refinement through continuous feedback from IT experts and end-users. Core features include mobile and web-based complaint submission, automated categorization and prioritization using supervised machine learning, geotagenabled reporting, workflow dashboards for staff and management, and role-based access control to ensure data security. Evaluation was conducted in two phases: IT experts assessed the system using ISO/IEC 25010 software quality standards, while Member-Consumer-Owners (MCOs) and cooperative personnel evaluated functional suitability, performance efficiency, usability, and overall acceptability. Results demonstrated excellent ratings across all quality dimensions, with end-users reporting faster complaint resolution, improved accessibility, and high satisfaction with the system’s usability. The findings confirm that AI-driven segmentation enhances complaint management efficiency, reduces delays caused by manual routing, and strengthens transparency and accountability in service delivery. Beyond improving daily operations, the CCMS supports compliance with government mandates such as Republic Act No. 11032 and NEA reporting requirements. Ultimately, the system fosters stronger relationships between cooperatives and their consumers by ensuring that complaints are systematically recorded, properly addressed, and resolved in a timely manner.

Stabilizing organic clay soils is pretty tough in geotechnical engineering because these soils are highly plastic, have low strength, and using common stabilizers like cement and lime often brings environmental concerns. This paper takes a close look at two new ways to stabilize soil using biological processes: Microbially Induced Calcite Precipitation (MICP) and Enzyme Induced Calcite Precipitation (EICP). This research looks closely at existing studies and experimental data to compare how well different sustainable options improve organic clay soil. The research uses supervised machine learning methods, like Random Forest, to build predictive models for soil stabilization. These models are based on key factors such as soil type, treatment concentration, curing time, and microstructural features. The results show that both MICP and EICP clearly improve the mechanical properties of soil. MICP can boost Unconfined Compressive Strength (UCS) by anywhere from 10% to 66% depending on the soil type, while EICP helps bring down the liquid limit from 79% to 58.It goes up by 8% and increases the plastic limit from 30% to 47.8%Putting biochar into MICP (MICP-BIN) really changed things, increasing shear strength by 389.It was 5% higher than soil that hadn’t been treated. Using SEM, EDX, and XRD to look at the microstructure, it was clear that calcium carbonate precipitation was the main way the soil got stabilized. The crystals form and clump the soil particles together, which reduces the spaces between them. The machine learning models were able to predict pretty accurately how effective the treatments would be. Looking at which features mattered most, it turned out that calcium carbonate content, curing time, and the soil’s initial plasticity were the key factors. This study offers a basic framework for choosing and improving bio-cementation methods based on data to stabilize cohesive soil. It focuses on a sustainable way to reduce carbon emissions while improving geotechnical performance in infrastructure projects.

The growing need for sustainable and lightweight building materials has accelerated research on alternatives to conventional concretes, out of which Lightweight Expanded Clay Aggregate (LECA) concrete has emerged as a promising solution. However, the high porosity and nonlinear mechanical behavior of LECA concrete complicate the accurate prediction of compressive strength through conventional empirical models. The main focus of the paper is on identifying a comprehensive machine learning-based framework for modeling and predicting the 28-day compressive strength of LECA-based lightweight concrete. The dataset was created and preprocessed by using statistical normalization and correlation analysis. In this study, five supervised machine learning models—Multiple Linear Regression (MLR), Support Vector Regression (SVR), Random Forest (RF), eXtreme Gradient Boosting (XGBoost), and Categorical Boosting (CatBoost)—were developed and fine-tuned using a grid-search strategy combined with ten-fold cross-validation. The quality of the prediction made by each model was evaluated by means of standard performance indicators, such as the coefficient of determination (R2), root mean square error (RMSE), mean absolute error (MAE), and mean absolute percentage error (MAPE). After the evaluation, the models were subsequently compared and ranked according to the Gray Relational Analysis (GRA) method. The comparative assessment shows that CatBoost demonstrated the most reliable performance, achieving an R2 of 0.907, RMSE of 3.41 MPa, MAE of 2.47 MPa, and MAPE of 10.05%, outperforming the remaining algorithms. To interpret the significance of features, SHAP (Shapley Additive exPlanations) analysis was applied, which identified water and LECA content as the dominant factors influencing compressive strength, followed by the cement and fine aggregate proportions. The findings reveal that the ensemble-based gradient boosting model is capable of capturing intricate nonlinear interactions, as observed in the heterogeneous matrix of LECA concrete.

ABSTRACT Rolling element bearing faults are a major source of failure in rotating machinery, making accurate and reliable fault diagnosis essential for condition monitoring and predictive maintenance. This study presents a supervised machine learning (ML) based framework for bearing fault classification using statistical features extracted from vibration signals. Standard preprocessing techniques, including feature extraction, feature standardisation, and label encoding, were employed to prepare the data for model training. Multiple classifiers were evaluated, and their performance was assessed using standard evaluation metrics and confusion matrix analysis. The results show that the Decision Tree (DT) classifier achieved the highest classification accuracy of 93.57%, indicating its strong capability to model nonlinear relationships between vibration features and bearing fault categories. The k-Nearest Neighbour (kNN) classifier recorded a slightly lower accuracy of 92.52% but achieved a high AUC of 98.60%, reflecting excellent class separability and generalisation capability. The strong AUC value suggests that kNN effectively distinguishes between healthy and faulty bearing conditions as well as among different fault categories when vibration features are properly scaled. Overall, the results indicate that while tree-based models offer higher raw classification accuracy, probabilistic and distance-based approaches provide strong generalisation performance and robust fault separability. The proposed approach shows strong generalisation capability, with only limited misclassification observed between certain fault types exhibiting similar vibration characteristics. Feature importance analysis further highlights the critical role of higher-order statistical features in distinguishing bearing health conditions. Overall, the findings confirm the effectiveness of the proposed framework for vibration-based bearing fault diagnosis and its potential applicability in real-world condition monitoring systems.

A systematic review of machine and deep learning techniques for acute lymphoblastic leukemia diagnosis.

This study aimed to develop and validate a machine learning-based model for predicting 30-day mortality in cardiac surgery patients and to implement a functional, clinician-oriented web application that enables the real-time use of the model. A retrospective cohort of 325 cardiac surgery patients was analysed using supervised machine learning. After preprocessing and clinical feature selection, several models were trained and evaluated through cross-validation. XGBoost achieved the best results, with an AUC-ROC of 0.968, recall of 0.800, and Brier score of 0.058. To facilitate clinical usability, a web-based application was developed using StreamLit, enabling clinicians to input patient data and predict mortality in real time. The application includes SHAP-based explainability for each prediction, thereby ensuring model transparency. Preliminary feedback from clinicians indicated that the tool was intuitive and informative and showed potential for preoperative risk assessment. The integration of a robust ML (machine learning) model with a functional clinical application offers a practical tool for supporting decision-making in cardiac surgery. This combined approach enhances both accuracy and accessibility, which are key to real-world impacts. Future work will involve multicentre validation and user-centred refinement.

ABSTRACT The expansion of irrigation in central Tunisia has reshaped landscapes, creating ecological conditions favourable to Phlebotomus sand flies and Leishmania transmission, thereby increasing leishmaniasis risk. To assess these dynamics, a geospatial classification and risk-indexing framework was developed that integrates remote sensing, supervised machine learning, and ecological reasoning on vector habitats. The analysis covered three Central Tunisian sites and their narrower environments – Gafsa, Kairouan, and Sidi Bouzid – from 1984 to 2020. Habitat clustering revealed marked spatial and temporal variability in land-cover composition. Low-risk habitats remained dominant, while irrigated high-suitability areas expanded progressively, particularly after the mid-1990s. Semi-natural habitats fluctuated, and very low-risk areas remained minimal. Ecotone analysis showed strong variability in transitional habitat edges, with peaks corresponding to increases in area-averaged risk indices. Territorial risk indices increased over time in all sites, reaching 0.470 in Gafsa (2012), 0.619 in Kairouan, and 0.507 in Sidi Bouzid (2020). These results indicate that irrigation-driven landscape changes enhance human – vector contact and transmission potential, underscoring the need to integrate irrigation practices into targeted, evidence-based vector management strategies.

The development of high-performance gas sensors is crucial for ensuring safety and efficiency in the emerging hydrogen economy, particularly for detecting hydrogen (H2) and ammonia (NH3), which are essential for hydrogen storage, transportation, and energy applications. Hydrogen is highly flammable, with a lower explosive limit of 4%, while ammonia is toxic and can cause severe health hazards; thus, their early and accurate detection is critical to prevent accidents and ensure safe handling. However, most hydrogen sensors exhibit cross-sensitivity to ammonia, making it challenging to distinguish between the two gases. Additionally, blends of ammonia and hydrogen are considered as alternative fuels to achieve zero-carbon emissions. Detecting them in mixture form is essential, as the flammability and toxicity limits of the mixture differ from those of the individual gases, requiring precise monitoring for safety, process optimization, and efficient fuel utilization. In this study, we employ palladium (Pd) nanoparticle-decorated electrostatically formed nanowire (Pd-EFN) sensor for the selective detection of H2, NH3, and their mixtures at low concentrations. The EFN sensor, a multiple-gate depletion-mode field-effect transistor (FET) fabricated using complementary metal-oxide-semiconductor (CMOS)-compatible processes, provides unique multigate electrostatic control, enabling enhanced sensitivity and selectivity. Experimental results demonstrate a highly reversible response, with distinct "electrostatic fingerprints" observed across different back-gate voltages, allowing for improved gas differentiation. Using supervised machine learning techniques including Linear and Kernel Support Vector Machine, AdaBoost, Gradient Boosting, Extra Trees, Random Forest, Decision Tree, Linear Discriminant Analysis, and K-Nearest Neighbors, we achieved up to 94% classification accuracy in distinguishing H2 vs NH3 and H2 vs (NH3 + H2), respectively. Additionally, adopting a transfer learning approach using the VGG-19 neural network and leveraging sensor response maps as inputs, further improved accuracy to approximately 97 and 96%, respectively. Furthermore, the ability to discern the individual gases and the mixture (H2/NH3/(NH3 + H2)) was improved from 77 to 87% with the use of transfer learning. The ability to selectively identify individual gases and their mixtures using a single sensor with high accuracy, without the need for sensor arrays, paves the way for advanced, miniaturized, and cost-effective gas sensing platforms, demonstrating potential for real-world applications in hydrogen safety and environmental monitoring.