Abstract Human pose estimation is a fundamental task in computer vision with widespread applications in human–computer interaction, sports analytics, and healthcare. While convolutional neural networks (CNNs) and Transformers have achieved notable success, they often struggle to capture structured body relationships, handle occlusions, and generalize effectively across diverse environments. Graph Neural Networks (GNNs), which represent human poses as structured graphs, offer a compelling alternative by explicitly modeling spatial and temporal dependencies among body joints. This survey provides a comprehensive review of GNN-based pose estimation approaches, encompassing spatial GCNs, spatiotemporal models, graph–Transformer hybrids, and hypergraph frameworks. We analyze these methods along key dimensions, including graph construction, learning paradigms, attention mechanisms, and computational efficiency, using standard benchmarks such as Human3.6 M, COCO, and MPI-INF-3DHP. Our review identifies several emerging trends and critical limitations. These include high computational cost, limited generalization to unconstrained scenarios, and inconsistent evaluation protocols. To advance the field, we outline future research directions, such as hybrid GNN–Transformer architectures, lightweight models for edge deployment, multi-modal fusion, and self-supervised learning strategies aimed at reducing annotation dependency and improving cross-domain robustness.

The proliferation of AI-generated medical deepfakes, such as tumor insertions or removals in diagnostic scans, threatens patient safety and healthcare integrity. Existing detection methods often lack robustness against adversarial attacks or fail to integrate multimodal feature representations. To address these gaps, we propose AFFETDS (Adversarial Feature Fusion Enhanced Tumor Detection System), a novel ensemble framework combining adversarial training, feature fusion, and weighted voting. AFFETDS leverages adversarial attack methods (PGD, FGSM) to harden the model, fuses high-level ResNet50 features with handcrafted HOG descriptors, and employs an SVM-based ensemble classifier. Evaluated on a curated dataset of 1378 MRI scans (774 real, 604 manipulated) from TCIA and ADNI repositories, AFFETDS achieves state-of-the-art performance with 91.5% accuracy, 90.7% precision, and 91.2% recall, outperforming baseline models (SVM: 86.2%, CNN: 88.4%). The framework's ROC-AUC (0.80) and calibrated confidence scores demonstrate superior generalization across diverse imaging conditions. The ability of combining the adversarial techniques with multimodal feature fusion, our proposed AFFETDS framework improves the detection of subtle tumor manipulations, presenting an important safeguard to maintain the authenticity of medical images. The findings of research work underscore the urgent need of proactive defenses against growing deepfake threats in healthcare applications.