Can AI revolutionise robotic surgery?

The rapid progress of Artificial Intelligence (AI) technologies in the last 5-10 years has led many to associate it with robotic surgery systems. Currently, however, few robotic surgery systems are equipped with AI-driven human-robot interaction capabilities.

AI offers numerous opportunities for the advancement of robotic surgery. It can facilitate interaction mediums between surgeons and surgical robots, for example by recognising surgeons' movements (e.g. head, eyes, hand) and converting them into an action command for the surgical robot.

AI can also enable verbal manipulation of a surgical robot through speech recognition arm. Although the precision and the accuracy of speech recognition has improved with the integration of deep learning in speech recognition, this type of technology remains in its early stages and requires further development to become reliable.

AI facilitates robotic instrument positioning. For example, ML algorithms in orthopedic surgery robots allow pre-operative planning by building a virtual model of the patient's anatomy and enable the creation of a trajectory for intervention (e.g. drilling, screw implantation). This reduces the chance of human error.

So, when will AI become widely implemented in robotic surgery systems? Currently, its use is restricted to image recognition algorithms for pre-operative planning. There is currently no clear path for other forms of AI in robotic surgery. Regulations are the biggest roadblock.

Regulatory frameworks are not built to accommodate adaptive technologies such as AI because AI algorithms constantly learn and change. When an algorithm adapts, it is no longer the same algorithm and cannot be utilised in the medical practice without updating approvals. While their understanding of AI remains vague, regulatory bodies view this unpredictability as too risky to approve for a surgical robot. They are in the process of designing new methods to regulate AI, but this will take years to come into effect.