Pinpoint accuracy in surgical technique has advanced beyond what many ever thought possible. Unfortunately, such refinements have been slower to reach the perioperative business front. OR leaders often rely on vague metrics and educated guesses to guide their teams.
Ambient artificial intelligence (AI) is changing that. When integrated with data from the electronic health record (EHR) and predictive algorithm software, advances in ambient AI tools are giving perioperative leaders far more accurate data to forecast room utilization, turnover rates, and patient throughput. For education, ambient AI is helping teams identify safety risks and learn from them. And ambient AI documentation technology is becoming a common tool to help clinicians, including surgeons, document more thorough and patient-focused workups.
But ambient AI is no panacea. Like any tool used in surgical care, applying it effectively requires knowledge, creativity, planning, and safety mechanisms, as well as the flexibility to learn and adapt.
The following case studies illustrate how ambient AI is being successfully adopted in major academic medical centers to drive operational improvements, enhance safety education, and enable more efficient communication with patients before surgery.
For almost 2 years, 28 ORs at Tampa General Hospital in Florida have been equipped with Apella—a computer vision technology that analyzes real-time video data as part of an ambient AI platform linked to the EHR, OR dashboard, and restricted remote logins.
This integration translates OR observations through AI-driven predictive modeling, providing a clearer understanding of what is happening—and what will happen—in the OR. It is helping leaders and frontline staff manage OR activities with greater precision. Everything from time spent in the OR to turnover efficiency, room cleaning proficiency, traffic patterns, and patient throughput is tracked.
How the ambient AI platform works
“The system is continuously learning, and the predictability it provides is helping us learn too,” says Kate Ulrich, MS, BSN, RN, NEA-BC, Tampa General’s senior vice president of perioperative services. In daily practice, Ulrich monitors live views of the ORs through a camera system she can access at any time. The system predicts schedule accuracy based on historical and real-time data, comparing these projections with EHR-generated calculations. This information appears on the OR’s main dashboard and can also be accessed via mobile login.
Ulrich can query the system’s data to answer specific questions—for example, she recently examined block utilization for a particular service line and, using predictive case durations, determined that 3,000 OR minutes could be freed for additional cases.
A benefit of Apella is its more accurate case-end predictions, including for complex cases involving multiple surgeons, such as breast reconstructive surgery. For patient throughput, a live view of each case allows preoperative nurses to better plan for case completion without needing to call into the room and interrupt the team.
Ulrich describes this real-time insight as a “game-changer” compared to the less accurate case-end estimates previously based solely on EHR data. The improved reliability also helps with staffing accuracy, as scheduled cases and add-ons can now be predicted within a normal workday. As a result, she no longer needs to ask staff to stay beyond their scheduled shifts, giving teams a clearer outlook on their day (sidebar, Metrics for ambient AI deep dive).
Apella also refines communication with surgeons by automatically sending text alerts when patients are wheeled into the OR, prepped, and draped. Since implementing these automated alerts, Ulrich’s team has seen fewer phone calls and greater situational awareness—two major wins for daily efficiency. “We were looking for a seamless, data-driven tool that enables efficient workflow patterns through real-time insights without disrupting clinical workflow—we’ve found it,” she says.
Building in AI safety precautions
Patient privacy was a top concern for Ulrich and her team before implementation. To address this, safeguards were built into the system governing what each team member can view from the ambient AI-generated data. Charge nurses, clinicians, anesthesia personnel, nurse managers, the business manager, and clinical directors all have access to various aspects of the data and predictions, but with customized security permissions. Ulrich and her team configured and refined these permissions as they adapted to the system.
Helping teams embrace the AI shift
Transitioning to Apella represented a significant change-management effort. Ulrich had to help OR staff become comfortable with being recorded. To do so, she enlisted experts from human resources, risk management, compliance, and surgical governance—including the chair of surgery, surgical chiefs, and committee leaders—to facilitate transparent communication and dialogue. Together, they created frequently asked questions and hosted multiple team meetings to allow staff to voice any doubts and receive clear answers.
A key message: the ambient AI platform would be used solely for efficiency and safety, not disciplinary action. “The nice thing about this system is that it produces an unbiased record of what happened with strictly factual data, which helps our clinicians see activity as it happens, without room to wonder or question,” Ulrich says.
As her team continues to work with Apella, new features and updates are offering additional opportunities. For example, Ulrich recently began using an automated report generated 90 days in advance to refine case scheduling. In most cases, the AI-predicted time is shorter than what had been scheduled.
At the Spae Lab at Duke University in Durham, North Carolina, lab director and assistant professor of surgery and biostatistics and bioinformatics J. James Jung, MD, PhD, FRCSC, FACS, is leveraging his experience with ambient video learning in the OR to build AI-enabled methods for measuring perioperative performance, safety, and operational efficiency. During his PhD training in clinical epidemiology at the University of Toronto, Dr Jung helped develop the foundational measurement framework using OR video recordings that became the OR Black Box® platform. One key finding from this early work was identifying how frequently auditory distractions—such as door openings or phone rings—occurred, and which intraoperative events they influenced. These insights informed post-case debriefs and educational modules.
Using ambient AI to improve outcomes
Today, Dr Jung’s team uses continuous, de-identified recordings of the intraoperative environment—including video, audio, and device data—to create a structured record of each surgical case. Authorized clinicians can review curated summaries of key practices, such as safety checklist adherence, with time-stamped annotations for review. One focus in the lab is applying computer vision to automate identification of intraoperative events like significant bleeding in real time, enabling earlier mobilization of resources to mitigate downstream consequences. The team is also researching automated assessments of both technical and nontechnical skills using validated metrics. Dr Jung’s goal is to understand how OR team performance correlates with patient outcomes, identifying opportunities for improvement and personalized coaching for trainees.
“When needed, we have the capability to dive deeper to assess intraoperative factors underpinning performance such as safety threats, resilience behaviors, and technical and nontechnical skills,” Dr Jung explains. He also collaborates with the Duke perioperative team on “Better Together” rounds, in which objective findings from analyzed cases inform multidisciplinary education for nurses, scrub techs, surgeons, anesthesiologists, and trainees. Recent topics have included preventing positioning-related injuries, optimizing surgical safety checklist adherence, and improving team communication during crisis management.
Engaging clinicians with ambient AI data
For clinician leaders seeking to implement ambient AI for education and quality improvement, Dr Jung recommends five strategies for success:
❶ Lead with a “just culture”
❷ Establish a clear, nonpunitive policy for performance assessment
❸ Measure teams, not just individuals
❹ Broaden participation by evaluating nontechnical skills such as communication, situational awareness, leadership, and resilience
❺ Create a multidisciplinary governance group to set safe, transparent data-access rules.
“Start small and review fast,” Dr Jung advises. Begin with a pilot in a few ORs, focusing on one or two concrete targets such as noise reduction during critical steps or instrument readiness. Then, learn from clinicians about the real pain points and use ambient technology to gather pertinent data. Follow up by sharing brief, actionable metrics with the team. “In our early experiences, high distraction rates were eye-opening and helped build engagement,” he notes.
Clinicians are also exploring the use of ambient AI for documentation. A recent study surveyed physicians—including those in surgical settings—at Mass General Brigham Hospital in Boston and Emory Healthcare in Atlanta to understand their perceptions of several types of ambient documentation technologies that capture clinician-patient conversations and generate structured notes using generative AI.
Ambient documentation technology use analyzed
The study’s primary goal was to evaluate whether ambient documentation tools help physicians reduce burnout and increase engagement during patient interactions. Survey responses revealed high satisfaction scores and supportive feedback, with one physician noting their assessment and care plan were “more complete” when using ambient documentation technology compared to traditional, manual note-taking.
However, some survey respondents pointed out workflow and technology limitations, including gaps in specialty-specific content and challenges with EHR integration that added time to documentation. These findings show both the benefits of ambient AI documentation and the ongoing need for tighter EHR integration and specialty-tailored customization to make these tools most effective.
Ethical and data-privacy considerations are central to ambient AI adoption. Staff concerns about surveillance and patient needs for privacy and informed consent must be thoughtfully addressed.
“In such a fast-evolving landscape, the need for clear guidance and cross-sector collaboration is more urgent than ever, especially when considering applications for patient care,” cautions Jonathan Perlin, MD, PhD, MSHA, MACP, FACMI, FAMIA, The Joint Commission’s president and CEO. “AI holds enormous potential, but we must encourage and support responsible AI adoption in a way that upholds the trust of patients, staff, and communities.”
Resources to advance AI safety
The Joint Commission recently released guidance on AI usage in healthcare, developed in collaboration with the Coalition for Health AI (CHAI). Dr Perlin says the guidance is designed to help health systems navigate the complex and rapidly evolving adoption of AI. “Recognizing the urgency and scale of AI’s potential impact on healthcare, this guidance provides a foundational roadmap for responsible implementation across the US healthcare system,” he explains.
Key focus areas in it include:
• Interoperability
• Ethical use of patient data
• Deployment strategies for first-of-its-kind AI initiatives.
To support these efforts, The Joint Commission and CHAI are also introducing AI governance playbooks, practical implementation tools, and voluntary certification programs. The goal, Dr Perlin says, is to “equip healthcare organizations with leading practices to advance AI responsibly, in alignment with the highest standards of patient safety, data ethics, and care quality.”
Across these initiatives, one theme stands out: ambient AI delivers value when used with intention. Whether it is to optimize case scheduling, support real-time education, or streamline documentation, success depends on clear governance, thoughtful implementation, and ongoing collaboration among stakeholders. As the above adopters show, ambient AI is not a replacement for human expertise, but an amplifier of it, offering OR leaders the data and insight they need to drive safer, smarter, and more efficient surgical care.
—Carina Stanton is a freelance science journalist living in Denver.