Sri Sai Vignesh Kadiyala

Developed an advanced fracture diagnosis classification system using a Hierarchical Vision Transformer (H-VT) with cross-modal attention in Python and PyTorch, integrating multi-modal fusion, CNNs, and transformer architectures for X-ray, CT, and MRI analysis. Implemented uncertainty quantification and explainable AI features (attention maps, feature attribution, interactive probing) to improve clinical trust and usability. Validated on a multi-centre dataset of 54,900 images, achieving 96.1% accuracy with multi-modal input and 92.3% with single modality, while reducing diagnostic time by 93%. Deployed using cloud and edge computing frameworks with Docker, Rubbernecks, and Flask APIs, ensuring scalability and accessibility in real-world healthcare settings, with 93% of outputs rated clinically actionable by specialists.