A load-bearing wall is a structural component designed to support vertical loads while resisting lateral forces acting parallel to its plane, such as those induced by wind, seismic activity, and earthquakes. These lateral forces can cause deflections, particularly critical in high-rise buildings, as they help prevent structural failure under seismic loads. Shear walls, commonly known as reinforced concrete (RC) structural walls, are crucial in enhancing a building’s resistance to earthquakes by acting as an effective lateral load-resisting system. The characteristics of the walls in a building have a strong influence on the overall structural performance of the building. Hence, a thorough understanding in the behavior of the walls under seismic loading is essential. Operational effectiveness of different structural arrangements such as conventional reinforced concrete (RCC) frames, arrangements using shear walls, arrangements using dampers, and combinations may be different to a significant extent when they are placed in seismic situations. This study examines how the position of shear walls in reinforced concrete constructions effect significant parameters like ground movement, base shear, and foundation forces especially in multi- storey structures. The effect that shear wall arrangement has on seismic behavior of such edifices is analyzed by a response-spectrum analysis which is based on seismic intensity. The analysis is done using the ETABS software with emphasis on the difference in bending moments, shear forces and axial loads, at varying shear -wall positions.
Federated learning has emerged as a promising paradigm for privacy-preserving healthcare analytics, yet its effectiveness in heart attack risk prediction is limited by non-IID client distributions, class imbalance, and subgroup bias. In this work, we present Fairness-Aware Federated Learning (FA-FEL), a continent-based client-level SMOTE-based stacked ensemble framework for heart attack risk prediction on the UNIBS Heart Attack Analysis & Prediction 2024 dataset. The proposed pipeline combines local preprocessing, client-wise SMOTE balancing, Optuna-tuned XGBoost–LightGBM stacking with logistic regression meta-learning, fairness-regularized local optimization, and fairness-aware federated aggregation. This design enables institutions to retain data locally while collaboratively learning a global predictor that is both accurate and equitable. Experiments show that FA-FEL achieves 98.92% accuracy, 0.9849 F1-score, 0.9991 ROC-AUC, and 0.0110 Brier score on the held-out test set, while maintaining low subgroup disparity across all continent-based clients. The results demonstrate that client-specific imbalance correction and fairness-aware federated optimization can substantially improve clinical risk prediction without centralizing sensitive EHR data.
An optimization-based approach to AI-driven personalized learning, aiming to address the challenges of traditional, one-size-fits-all educational systems is presented in this article. It proposes a model that dynamically adjusts learning pathways based on real-time data from learners, enabling continuous customization of instructional content, assessments, and feedback mechanisms. The model optimizes various educational parameters to maximize learning outcomes, engagement, and retention. The results show that AI-powered personalization significantly improves student performance, engagement, and retention rates. However, challenges like data privacy, algorithmic bias, and scalability remain.
Ocular diseases are a global public health concern, demanding precise diagnostic methods. This study tackles the challenge of accurate recognition from low-resolution ocular images using a combined approach of Super-Resolution Generative Adversarial Networks (SRGAN) and Convolutional Neural Networks (CNN). Our method involves enhancing image quality through SRGAN to create high-resolution versions of low-quality ocular images. Subsequently, a CNN-based recognition model is trained on these enhanced images, learning discriminative features for precise classification of various ocular diseases. Performance evaluations on the Ocular disease dataset demonstrate the practicality and effectiveness of our approach. Beyond diagnostics, this SRGAN-CNN combination holds potential for telemedicine, enabling swift and reliable eye disease diagnosis, especially in remote areas. This synergy not only improves diagnostic accuracy but also streamlines analysis, enhancing treatment efficiency and patient outcomes.
Ovarian cancer is often detected only in advanced stages. Early symptoms are either absent or very mild, which makes them difficult to notice. Patients usually fail to recognize these signs, and primary care doctors also face difficulty in separating them from other common conditions. This study looks into how machine learning, can help find ovarian cancer early through blood tests. CatBoost algorithm is used to construct a predictive model on a dataset containing 48 blood biomarkers. Key findings from this work include that the top 18 most important features being discovered by removing redundant features using the MRMR feature selection. The data underwent correlation analysis to know the interrelations between features of importance and target variables. CatBoost model with MRMR feature selection showed outstanding performance of accuracy of 97.7
