A
recent qualitative study identified potential barriers and facilitators to
implementing AI-based interventions in surgery. The authors suggested that these
findings can be aligned with standard implementation strategies to better align
the expectations of AI-based interventions in surgery with their actual
capabilities. This approach establishes a foundation for more successful
implementation, thereby increasing the likelihood that surgeons will
effectively utilize this new technology to enhance patient outcomes.
This
qualitative study is published in January 2026 in JAMA Surgery
Applying
Artificial Intelligence (AI) Potential in the Operation Theatre (OT) Room
Artificial
intelligence (AI) is a transformational tool for surgery that improves
outcomes. However, the gaps between expectations and reality can hinder
implementation. Given the high costs of implementing AI, understanding barriers
and facilitators is critical to developing implementation packages and
strategies that optimize the likelihood of successfully delivering and
sustaining AI interventions that improve perioperative outcomes in surgical
patients.
The
Operating Room Black Box (ORBB) is a proprietary product developed by Surgical
Safety Technologies and serves as a valuable case study for examining the
impact of expectation gaps on the implementation of intraoperative AI
interventions. ORBB is among the pioneering programs that utilize ambient AI
within the operating room to capture and process audiovisual data related to
team communication and performance, alongside intraoperative patient data.
Evidence shows that ORBB can improve outcomes by identifying errors, enhancing
team performance, ensuring safety, and providing training data.
Study
Overview
This
qualitative study was conducted at 3 large academic centres via semi-structured
interviews with surgeons and implementation leaders of the AI intervention to
identify barriers and facilitators to implementing AI–based interventions that
improve intra- and postoperative care. Through a screening survey, 30 surgeons
and 17 implementation leaders from 3 centres that implemented the AI
intervention were interviewed. The intervention consisted of surgical video
content management, which builds and stores intraoperative videos for
evaluation; surgical case review, which automatically flags cases for review
based on outlier identification or user request; and surgical safety checklist
compliance, which automatically rates adherence to intraoperative safety protocols.
The primary outcome was misalignment between participants’ expectations of the
AI intervention technology and the programme’s deliverables.
Key
Findings
In
this study, 57% of the surgeons maintained a neutral perspective on the
technology, 37% expressed favourable opinions, and 7% expressed negative views.
Interviewees
identified the following 4 major themes that highlighted misalignment between
user expectations and the experience of using the technology
The
AI model needed considerable additional training to be usableAccessing
data on surgical cases was difficult and time-consumingThe
program showed limited ability to predict postoperative complicationsThe
program generated a few academic deliverables
Potential
Learnings for Stakeholders
AI-derived
data may not be immediately actionable, and slow turnaround times can limit its
use in real time. To adapt AI models for healthcare settings, the
implementation team must be prepared to reduce frustration and abandonment.
Small pilot projects allow users to provide feedback that improves
implementation and acceptance, because users can shut down the programme at
minimal cost if it proves ineffective. For academic surgeons, AI may be more
valuable for studying team communication and training than for studying rare
complications. Other common barriers to implementation include poor workflow
integration, limited stakeholder engagement, and insufficient institutional
support. These challenges particularly affect smaller hospitals with fewer
resources. Less-resourced institutions should consider implementing proven,
ready-made programmes rather than adapting interventions and should focus their
resources on technical support.
Reference:
Thornton M, Cher
BAY, Macdonald C, et al. Expectations vs. Reality of an Intraoperative
Artificial Intelligence Intervention. JAMA Surg. Published online
January 14, 2026. doi:10.1001/jamasurg.2025.6029