October 27, 2025

Building trust in AI, making it work for perioperative nurses

By: Brita Belli
Share

Artificial intelligence (AI) has made inroads into nearly every area of healthcare. With nursing shortages continuing—marked by the loss of some 100,000 nurses following the COVID-19 pandemic and projected deficits of 20% or more in some states—AI-based tools that improve access to information, streamline efficiency, monitor patients, track procedures, and aid administrative processes can be a lifeline for overextended nursing teams.

Yet introducing these tools into daily workflows—and even into the OR—has been met with equal parts enthusiasm and skepticism among nurses.

Janice Kelly, MS, RN, NI-BC

“Nurses will question everything,” says Janice Kelly, MS, RN, NI-BC, president of AORN Syntegrity and a board-certified informatics nurse with more than 25 years of experience. “They fear that AI could replace them. Nursing relies heavily on evidence-based practice and tradition. We make decisions using knowledge and critical thinking, and AI is a disruptor to this approach.”

Kelly is committed to helping nurses see how AI can serve as “a new team member” with quick answers to practical questions, from locating supplies to confirming how to correctly position a patient in reverse Trendelenburg. She notes AORN eGuidelines+ now includes an AI assistant that allows nurses to type conversational questions and receive instant, evidence-based answers sourced directly from the guidelines—covering everything from positioning to skin prep, anesthesia, medication, handover tools, and room setup.

Beth Kaylor, RN, CAPS, CSPO

“Conversational AI is the biggest opportunity in the OR,” Beth Kaylor, RN, CAPS, CSPO, AI product manager and founder of the newsletter AI in Nursing, says. “Instructions for use [IFUs] and equipment manuals are essential, vast, and rarely light reading. These documents must be accessible for compliance with multiple regulatory agencies, but finding the right one when needed can be a challenge. With conversational AI, instead of hunting through documents and keyword searches, you can start a conversation with a bot that can access the manual’s content.”

Kaylor began her nursing career in the OR supporting new medtech and transitioned fully into technology by 2000. Today, she works at the intersection of AI and clinical work, developing OR-focused products, consulting with product teams to align AI tools with patient safety and user needs, and educating nurses on AI benefits and considerations via her newsletter/seminars.

“I see incredible potential, but I also hear the very real AI anxiety from nurses,” she says. “Their concerns aren’t just about job security. They’re worried about patient safety and clinical accuracy, and how these tools could impact care. Nurses need to be central in shaping how AI is developed and applied, so it strengthens care while protecting both patients and the professionals who care for them.”

 

Ramping up AI adoption

AI in hospitals dates back nearly five decades. In the mid-1970s, researchers at Stanford developed MYCIN, one of the first AI systems designed to identify bacteria causing infections and to recommend antibiotics. By the 1980s and 1990s, AI was being used for image analysis, detecting anomalies or assisting in measurements of X-rays, CT scans, and MRIs. By the 2000s, larger datasets, greater computing power, and improved algorithms accelerated progress dramatically.

Today, nearly all leading hospitals have integrated AI tools to streamline operations.

Scheduling and administrative workflow software. One example is LeanTaaS and its iQueue software, which is used in more than 1,000 hospitals across 180+ health systems. The promise includes increased surgical capacity, improved room and block utilization, and enhanced bed turnover rates and patient flow. In addition to software, hospitals receive dedicated support from the LeanTaaS team. Per the company, the results have been significant, including:

• an average of 30–50 additional cases per OR per year, representing roughly $20,000 per infusion chair and $10,000 per inpatient bed in additional revenue

• 7% higher staffed room utilization and 5% higher OR block utilization

• 30% shorter patient wait times, 50% less staff overtime, and a 20% increase in bed turns.

The range of available AI solutions is expanding rapidly. In 2024, Microsoft launched a healthcare agent service in its Copilot Studio, allowing hospitals—including early adopter Cleveland Clinic—to build custom AI tools from prebuilt healthcare templates. These tools can support a wide range of tasks, from scheduling and triage to patient communication.

Similarly, Luma Health offers AI-driven solutions to manage high call volumes and manual fax processing, which are two sources of delay and staff frustration. Its EHR-integrated platform enables automated faxing and provides a “patient-facing voice AI concierge” that can handle appointment details, refill requests, and preop prep questions.

The data processed before, during, and after high-stakes surgical procedures are staggering. When AI systems combine data from cameras, sensors, and patient records, they can deliver real-time insights that prevent errors and improve outcomes.

“Think about the cognitive burden of OR nursing,” says Kaylor. “It’s not just patient care, it’s managing all the resources that make it successful. There’s such a vast amount of information to learn. I’ve worked with hospitals that have 50,000 items in an inventory file. Now, we have that information at our fingertips through AI.”

Reducing costly errors. With so much to track, mistakes are easy and costly. AORN Syntegrity’s standardized procedure list includes over 2,700 surgical procedures across specialties, helping hospitals maintain consistent terminology for clearer communication among teams. MedCom Solutions, in partnership with AORN, applies AI to streamline insurance coding processes using those standardized terms.

Bill Hunt

“Our software ensures hospitals use the right terminology to get proper authorization,” explains Bill Hunt, president and CEO of MedCom. “It also helps ensure billing is processed correctly.” He offers an example: “If a partial knee replacement becomes a full replacement during surgery, the scrub nurse can notify the OR clerk to update the insurer. If that update is submitted within 24 hours, ideally the same day, the insurer will approve the upgrade. Wait longer than 24 hours, and it’s often denied.” These denials can cost hospitals tens of thousands of dollars per case, Hunt adds.

In 2023, US hospitals spent nearly $20 billion appealing claim denials, according to a Premier study, despite more than half of those claims being payable at submission. As the 2024 Costs of Caring report from the American Hospital Association (AHA) notes, “These denials are particularly concerning because they often occur for medically necessary care, which can result in direct patient harm.” Even when overturned, hospitals are rarely reimbursed for the administrative costs of appealing claim denials.

 

Involving staff in development, rollout

Ideally, hospital staff are involved in designing and implementing new AI tools, as innovators, advisors, and early adopters. They understand the workflow challenges firsthand, and their buy-in is essential to success.

A leading example is Teodor Grantcharov, MD, PhD, associate chief quality officer for innovation and safety at Stanford Health Care and professor of surgery at Stanford University, who developed the OR Black Box®, an AI-driven solution inspired by aviation’s flight data recorder. The system, combining audio, video, and AI analytics, is now used in more than 50 hospitals across the US, Canada, and Europe, including Stanford, Duke University Hospital, Brigham and Women’s, and Mount Sinai Health System.

Research using OR Black Box® data shows how often distractions occur. One study found OR doors opened a median of 42 times per case—roughly once every 2 minutes. Such distractions can impede surgical performance and increase the risk of errors, which simulated studies have shown can reduce speed, accuracy, and technical proficiency.

The OR Black Box® processes more than half a million data points per OR each day, tracking patient vitals, equipment performance, and patterns to:

• improve surgical performance

• support safety checklists

• optimize scheduling/room turnover

• enhance education and quality.

“If we really want to take surgical care to the next level, we need to embrace the technology that exists,” said Douglas S. Smink, MD, MPH, chief of surgery at Brigham and Women’s Faulkner Hospital, in a statement to Surgical Safety Technologies, the company behind the OR Black Box® platform. “We need to open up the [OR], the place that’s been…very secretive for decades, and use video and audio analysis to try to improve how we deliver care.”

 

Better handoffs, risk prediction

Roughly one-third of medical errors and adverse events happen during handoffs and care transitions. Standardized postoperative checklists often miss patient-specific risks, which is an area where AI can provide valuable support.

Joanna Abraham, PhD, FACMI, FAMIA

Joanna Abraham, PhD, FACMI, FAMIA, director of the Center for Applied eHealth Informatics and associate professor of anesthesiology at Washington University, has been developing AI-driven decision-support tools that predict postoperative complications such as acute kidney injury, delirium, DVT/PE, and pneumonia. Her work leverages AI insights to improve coordination and communication during the perioperative care continuum.

“I was interested in understanding the gaps and failures in the handoff process,” Abraham says. “Beyond the exchange of information, there needs to be an exchange of critical thinking—reasoning about potential complications and their root causes so the postoperative care plan is more personalized. How was a decision made in that plan?”

Her team’s tool summarizes a patient’s data and compares it with a larger patient group to generate risk predictions that support contingency planning based on modifiable factors.

“This tool is designed to encourage conversations about postoperative risks and guide care teams to ask the right questions to support continuity of care and enable proactive interventions before complications arise,” Abraham says. “Traditional handoff tools are mostly checklists and often lack explicit guidance on critical elements. Having the risk report embedded in the handoff tool forces the conversation.”

It is also especially helpful for nurses, she adds, who are responsible for spotting early signs of risk and delivering timely interventions.

 

AI-first hospitals

Some hospitals have aggressively integrated AI. Hartford HealthCare in Connecticut, for example, established the Center for AI Innovation in Healthcare, one of a few such centers nationwide. The system collaborates with Dimitris Bertsimas, associate dean for the Master of Business Analytics at MIT, to translate research into practice.

Bertsimas founded Holistic Hospital Optimization (H2O) to help providers improve patient flow using machine learning. H2O tools—developed at MIT and deployed at hospitals like Hartford HealthCare—include applications to manage patient length of stay, predict clinical deterioration, optimize nurse scheduling, and plan OR block scheduling based on historical data.

Implementing these tools has yielded measurable improvements, including shorter lengths of stay through proactive discharge planning and reduced nurse turnover by improving fairness in shift assignments. Recognizing the need for education, Bertsimas also offers a nine-lecture course at Hartford HealthCare to teach analytics methods to physicians, nurses, and administrators.

In 2024, Mount Sinai announced the opening of the Hamilton and Amabel James Center for Artificial Intelligence and Human Health, a state-of-the-art facility among the first of its kind at a US medical school. Mount Sinai even operates its own supercomputer, Minerva, in use since 2013.

As Abraham emphasizes, continued education is critical. “Clinicians think AI is taking over the world and they will lose their jobs. That’s why education and awareness are so important,” she says. “This is an additional tool, and it can offer another perspective.”

AI tools are here, but their success ultimately depends on human trust. When nurses are engaged early in design, implementation, and evaluation, AI becomes less of a disruptor and more of a partner. The technology has great potential, but true breakthrough will come when nurses view AI not as competition, but as a collaborator—one that amplifies their expertise and helps them focus on what matters most: safe, high-quality patient care.

—Brita Belli is an award-winning writer with published stories in The New York Times, National Geographic, MSN.com, and Alternet.

 

References

2024 Costs of Caring. America’s Hospitals and Health Systems Continue to Face Escalating Operational Costs and Economic Pressures as They Care for Patients and Communities. AHA. May 2024.

Abraham J. Improving Safety in Postoperative Handoff Communication with Telemedicine and Machine Learning. Agency for Healthcare Research and Quality. 2023.

Auerbach D I, Buerhaus P I, Donelan K, et al. A Worrisome Drop In The Number Of Young Nurses. Health Affairs. April 13, 2022.

Feijo S. Q&A: A roadmap for revolutionizing health care through data-driven innovation. MIT. May 5, 2025.

Innovator Awards Vendors Division: Semi-Finalist: LeanTaaS. Healthcare Innovation. March 6, 2024.

Jung J J, Juni P, Lebovic G, et al. First-year Analysis of the Operating Room Black Box Study. Ann Surg. 2020;271(1):122–127.

Landi H. Microsoft rolls out new healthcare AI tools in partnership with Epic, Paige.ai, major health systems. Fierce Healthcare. October 16, 2024.

Luma Health Announces Spark, Multi-Model Generative AI to Tackle Staff Work and Get Patients the Care They Need More Efficiently. PR Newswire. October 15, 2024.

Mount Sinai Opens the Hamilton and Amabel James Center for Artificial Intelligence and Human Health to Transform Health Care by Spearheading the AI Revolution. Mount Sinai. November 25, 2024.

Nursing Shortage Fact Sheet. American Association of Colleges of Nursing. Updated May 2024.

Join our community

Learn More
Video Spotlight
Live chat by BoldChat