Wrist fractures are a common problem in practice and emergency clinics, particularly in physically active individuals and the elderly. Current diagnostic approaches, dependent on trained radiologists and X-ray images, are slow and inhomogeneous. Motivated by advances in deep learning and the You Only Look Once (YOLO) family, our solution introduces FractureNet, a pruned and quantized YOLOv11-M version tailored for both accuracy and computational complexity in mobile fracture detection. We developed an Android app empowered with Artificial Intelligence for interpretable and accurate wrist fracture detection on the GRAZPEDWRI-DX dataset. Key innovations include high-performance detection scores with a mean average precision at 50 of 0.837, model compression reducing floating-point operations to 14.9%, and TensorFlow Lite optimization for low-resource deployments. Realized in offline form, it achieves an inference time of 1.18 s on the Samsung Galaxy S21, achieving fast and efficient detection. Streamlining fracture diagnosis, FractureNet educates patients for self-assessment, aids clinicians in practice clinics, and serves as a learning tool in medical education. Our contribution is a major step toward convenient, efficient, and reliable mobile healthcare solutions.