We have worked on multiple surgical ecosystems that involve advanced guidance technology, AI, and robotics. All of these products were started with user needs and market opportunities, but workflows, feature interactions, and system structures were typically determined by the most efficient way to use the technology, which leads to adoption barriers. Surgeons perceive the technology as complicated and are overwhelmed by the amount of change. The objective of this project was to identify the tension points between robotic solutions and natural behaviors in spine surgery. We aimed to conceptually design a robotic-assisted system that illustrates how a human-centered artificial intelligence (HCAI) surgical robot would differ from existing solutions.
Exploring Human-Centered AI: a Surgical Robotics Use Case
AI
Robotics
If an HCAI-based surgical robot experience was designed today, how would that look different from the current solutions?
Robotic-assisted spine surgery (RASS) is an emerging field with significant growth potential. However, despite promising research outcomes, RASS solutions remain underutilized due to concerns about complexity, their usefulness, ease of use, and behavioral control.
By minimizing disruptions to current freehand and navigated procedures, we aimed to seamlessly integrate robotic steps into existing surgeon-driven workflows, facilitating a smoother transition to robotic solutions, improving access to system benefits.
Procedures are optimized for robotics, not people. Change is necessary when adopting new technology, but change done strategically can avoid overwhelming surgeons and surgical staff, increasing adoption and promotion rates.
How We Did It
Common surgeon reported & observed barriers in current solutions
Current barriers to the adoption of surgical robots are an intimidating learning curve, a lack of trust in the technology, and a perceived lack of control for the surgeon. Future advancements will differentiate new robotic systems by minimizing changes to technical requirements and adding value only where necessary. To build trust, the AI should explain the logic behind its recommendations, and the robot should be reactive and context-aware rather than prescriptive, giving the surgeon greater autonomy.
Insights from observations
Robotic-Assisted Spine Surgery introduces a planning step to optimize the workflow: plan the placement of all screws before the patient is in the room. Once the patient is prepped and registered, the surgeon is supposed to align the robot to the pedicle, using the positioning to determine the incision. The problem is that this process removes tactile feedback from screw planning, which is critical to the surgeon’s decision-making and confidence in the plan. As a result, surgeons typically have their sales reps complete the planning on the robotic system, and they default back to their preferred use of fluoros and physical landmarks to approve or request adjustments to the plan. Some surgeons revert to placing screws by hand after using the robot for K-wires, prioritizing a sense of control and flexibility.
The challenge of adopting RASS technology stems from its steep learning curve and significant workflow changes. Surgeons often bypass new features because they prefer the control of manual methods. Strategic, gradual implementation is necessary to avoid overwhelming surgical teams, helping them fully utilize the robot's benefits while maintaining a comfortable sense of control and familiarity.
AI to Enhance Surgeon’s natural interactions
Our conceptual solution focused on supporting the current workflow to minimize the learning curve and build surgeon trust. Instead of a rigid, prescriptive plan, the robot offers real-time direction without removing the surgeon's sense of control. The primary changes aim to support the current workflow and minimize the learning curve for surgeons. By explaining AI recommendations clearly, trust is built, allowing surgeons to feel in control with real-time direction, rather than executing a rigid plan. The planning phase is reordered so that surgeons can use natural touch and physical space cues for trajectory assessment, while explainable AI highlights pedicle measurements to further reinforce trust.
Outcomes
The outcome is a system that becomes more intuitive and comfortable for users, making it easier for them to integrate into their organization. With the value of the system becoming more visible, surgeons can make decisions with greater confidence. The improved learning curve also fosters trust in automation, driving more seamless interactions between the surgeon and the technology.
