Abstract:
Reliable situation recognition in the operating room (OR) is the key aspect of the targeted support of the surgical team. Using intraoperative sensors and information about surgical processes, a situation recognition system (SRS) can recognize the current situation in the OR and provide this knowledge to context-aware systems (CAS). Current approaches for situation recognition focus on specific use cases and are not designed for transferability. However, due to the highly dynamic surgical environment, intelligent systems must be able to adapt to changing conditions. An SRS must therefore not only be applicable to specific surgical settings but also transferable to other scenarios.
A transferable SRS could enable broadly applicable CAS to flexibly support the surgical team in different surgical settings and thus be used scenario-independent. Based on an iterative development process, a comprehensive concept of a modular system architecture with four layers was developed and a framework prototype was derived to achieve fundamental transferability. The SRS can incorporate data from different sensors and uses type-based modules to process the information to support sensor-independent situation recognition. Via surgical process models, process knowledge is integrated and used as a crucial component for realizing intervention-independency. New approaches for process formalization based on BPMN and CMMN enable the mapping of standardized and non-deterministic process parts and thus the flexibility required to represent the highly variable intraoperative processes. By combining sensor and process knowledge, the SRS can recognize different situation knowledge, such as surgical phases. To realize broadly applicable CAS, the recognized situation knowledge is provided via an SDC interface. Using a standardized publish-subscribe pattern, various CAS, such as the OR-Pad for visualizing clinically relevant information, can subscribe to the context information and trigger context-aware behavior for the targeted support of the surgical team.
The results of the functional and argumentative evaluation proved the feasibility of the scenario-independent applicability and fundamental transferability of the SRS. By focusing on compatibility, maintainability, and portability during system design, several scenarios can be supported by the SRS. Moreover, the effort required to incrementally expand the functional complexity and variety of scenarios can be reduced to meet further challenges of transferability iteratively. Flexibility and easy adaptability characterize the overall system and enable it to cover deviating and new surgical settings. The SRS stands out from state-of-the-art approaches as it was realized with a focus on transferability required in the intraoperative area and promises broad applicability for supporting CAS. The results can be used as a basis for the development of a transferable SRS and also contribute to generally improving the transferability of intelligent systems.
In conclusion, this work demonstrates the development of a fundamentally transferable SRS for the broad applicability of CAS. The modular system architecture, the adaptable framework prototype, the BPMN- and CMMN-based process formalization, as well as standardized information provision for CAS, such as the OR-Pad, address the transferability necessary for the dynamic surgical environment and can contribute to the realization of intelligent ORs in the long term, in which surgical assistance systems support the intraoperative processes and the surgical team in a targeted and scenario-independent manner.
Situation recognition system