Abstract We present the neuro-symbolic anatomy engine, a novel artificial intelligence architecture that integrates structured anatomical knowledge with neural pattern recognition for comprehensive human anatomy modeling. Rather than focusing on individual procedures, the anatomy engine represents a Copernican shift: it encodes a universal model of human anatomy from which any surgical procedure can be simulated. This hybrid approach combines symbolic reasoning over anatomical relationships with neural network processing of text, images, and medical data. It enables transparent, traceable decision-making suited to surgical and educational applications. The system addresses critical limitations in current surgical training and has the potential to support lifelong learning, expand access to complex procedures, and drive surgical innovation.
Benz et al. (Tue,) studied this question.