Introduction

In times of labor and qualification shortage, increasing the efficiency in the operating room (OR) for patient safety plays a crucial role [1]. Among these efforts, the rapid provision organization of instruments is essential for the success of critical medical interventions. To manage this situation the assistance of systems based on artificial intelligence (AI) and robotics is envisioned.

Methods

AI is utilized in image recognition to determine geometry, type, number, orientation, and position of medical instru- ments and consumables. An innovative application of AI is presented, particularly using the YOLO (You Only Look Once) algorithm in combination with a camera system, to enable precise identification and distinguishing of surgical instruments.

Results

The YOLO-based camera system allows real-time detection and classification of instruments, capturing not only the type of instrument but also its number, orientation, and position. As a result, a voice-activated robotic system can pro- vide the required instruments to the surgeon upon request, thereby optimizing the surgical workflow. Furthermore, the camera observation system serves as an intra- and post-operational quality assurance tool by documenting the loca- tion/whereabouts and usage of instruments.

Conclusion

The presented AI approach provides significant contribution towards a prospective OR robotic assistance system con- trolled by a YOLO-based camera system, thus enabling efficient instrument provision to the surgeon. This technology may counteract OR labor shortage and enhances safety and efficiency in the OR. This contributes to transparency and traceability, augmenting compliance with necessary OR standards.

Based on current demographic developments it is predicted that in the near future (in the year 2035 and beyond ) more

than 500,000 positions in the inpatient, outpatient and intensive care cannot be filled in Germany. Similar developments

are known in other European countries.

In the past decades autonomous robots and mobile platforms to compensate the lack of manpower have been developed

and established, specifically for use in safe scenarios as e.g. within assembly lines or production, but also for the autono-

mous transportation of materials. Simple versions of such systems can already be found in domestic environments, as

e.g., in the form of vacuum or lawn-mowing robots, industrial cleaning robots or – in simpler forms –CNC milling ma-

chines, 3D printers or programmable kitchen machines.

Even though "care robots" have been the focus of various national and international R&D projects in recent years, there

has not yet been an urgent need to integrate them more closely into care processes. Reasons for this have been (a) the lack

of investment and business models due to the fact that nursing staff are still available at short notice, (b) a lack of infra-

structure such as comprehensive WLAN coverage, flat floors or closed doors in care facilities, day clinics or home envi-

ronment for zje self-localization and navigation of the robots, and (c) fixed processes and less flexible workflows in the

affected (care) facilities (intensive, inpatient, outpatient), which make it difficult to integrate mobile systems.

Introduction

Given the challenges of demographic change and staff shortage, surgical training plays a key role in sustainably provid-

ing high-quality surgical care. The assessment of cognitive load during training holds the potential to optimize and indi-

vidualize curricula and enhance learning outcomes. However, the development of the cognitive load during the curricu-

lum as well as the impact of surgical techniques, e.g. robotic or laparoscopic surgery, on the cognitive load, remains

unknown. In this study, we evaluate the differences in cognitive load between training in robotic surgery and laparo-

scopic surgery as well as the cognitive load development during the curriculum.

Effective and intuitive human-robot interaction requires interfaces that enable individualized support by adapting to hu-

man states and intentions. Such interfaces even enable continuous learning from human. For example, exoskeletons can

be used to compensate for movement disorders (1-4). Here, muscle activity recorded as electromyogram or the electro-

encephalogram (EEG) can be used to assist as needed (5), to infer movement intention (7) or to recognize subjective

failure in assistance (8). We call our approach of biosignal analysis that is embedded into the control of a robot and uses

the EEG and other biosignals embedded brain reading (6).