Surveys hardware-rooted trust mechanisms for AI-enabled IoT and edge computing systems, synthesizing literature on PUFs for device authenticity and AI model integrity under strict resource and security constraints.

A streamlined patient-specific workflow integrating deep learning-based segmentation with the GIBBON library and FEBio, fundamentally transforming the FEA preprocessing pipeline and enabling rapid, accurate lumbar spine biomechanical simulations from medical imaging data.

An automated FEA methodology for lumbar spine biomechanics integrating deep learning segmentation with computational modeling, streamlining the workflow from medical imaging to patient-specific spinal load distribution and stress simulation.

Evaluates Naive Bayes, KNN, SVM, LDA, Decision Trees, and DNN classifiers using bag-of-words and TF-IDF feature extraction for SMS spam detection, benchmarking LLM-enhanced approaches for robust, context-aware cybersecurity classification.

Presents a multi-objective optimization framework for water resource allocation addressing soil salinization and desertification through surface runoff management in watershed systems, with applications to mitigating environmental catastrophes from salt lake desiccation.

A hybrid of 12 feature extraction methods benchmarked against six classical ML classifiers for Alzheimer’s disease severity staging. A deeply supervised adaptable neural network achieves superior classification for neurological disease grading.

Benchmarks SAM and U-Net for breast tumor segmentation in both ultrasound and mammographic images. U-Net outperforms pretrained SAM for irregular shapes, indistinct boundaries, and high tumor heterogeneity, highlighting the importance of task-specific architectures in medical imaging.

Evaluates SAM and U-Net for concrete crack detection in bridges, buildings, and roads. SAM excels at longitudinal crack identification; U-Net is superior for spalling detection. A combined approach achieves the best overall performance for infrastructure safety assessment.

Explores deep learning-based segmentation for digital twin terrain modeling using USGS data to build global terrain and altitude maps for coastal regions, encoding land height and elevation modifications with precision across 5,000 worldwide segments.

Examines the emerging integration of artificial intelligence and finite element analysis in spine surgery, connecting data-driven clinical insight with patient-specific biomechanical simulation to support diagnosis, planning, and precision care.

Investigates associations between climatological parameters and the early spread of COVID-19 across Iran, supporting environmental analysis of infectious-disease transmission.

Applies convolutional neural network methods to chest X-ray images for COVID-19 diagnosis and localization of infected lung tissue.

Reviews and organizes AI applications in digital twin models for stormwater infrastructure, with a focus on resilient and data-driven smart-city systems.

Presents a hybrid artificial intelligence approach for forecasting economic growth from influential socioeconomic indicators.

Reviews physics-informed machine learning for computational medical imaging, connecting data-driven models with governing physical principles.

DQRE-SCnet presents a combined Deep-Q-Reinforcement Learning and Spectral Clustering approach for intelligent device selection in federated learning, reducing communication rounds while maintaining model accuracy across non-IID distributed data.

Uses gene expression programming and sensitivity analysis to examine parameters associated with gastric cancer tumor size and location.

Integrates finite element analysis with physics-informed neural networks for biomechanical modeling of the human lumbar spine.