Multi-agent Approximation of User Behavior for AR Surgery Assistant

The paper presents a new solution for a problem of adaptive data visualization in augmented reality interfaces used for intelligent surgery assistance. It is proposed to implement a multi-agent architecture capable of capturing the behavior and attention of users by formalization of focus and context, implementing the model of focused visualization and adjusting the virtual part of surgery scene in real time. The term of multi-agent approximation is introduced to formalize the approach. Software and hardware implementation is used for preoperative planning, 3D imaging, and surgery navigation. The developments were successfully probated at clinics of Samara State Medical University. This paper describes the details of the proposed solution and its implementation in practice.