TL;DR: Conversational AI has the potential to transform the operating room by providing instant access to critical information, automating policy creation, and enhancing patient support. From navigating complex manuals and regulatory documents to offering real-time vendor knowledge and creating collaborative knowledge bases, AI bots streamline processes and improve efficiency. These technologies empower healthcare providers to leverage AI at their own pace, offering personalized insights and support to both staff and patients, ultimately enhancing overall healthcare delivery.
While Siri and Alexa have been around for years, newer conversational AI models like ChatGPT and Claude AI are opening up exciting possibilities, especially in healthcare. Imagine having a super-smart digital assistant that’s available 24/7, ready to chat and help you out. It’s like texting an expert who can quickly tap into a vast library of information instantly. Customized AI bots can be trained to understand your messages naturally and can sift through relevant data to find what you need in seconds. Whether you’re looking for industry best practices or specific details about medical devices, they can pull up valuable info fast, making it easier to find and understand complex information that might otherwise be hard to track down.
While there are already many incredible advancements in AI for healthcare and exciting developments on the horizon, conversational AI could allow hospitals to leverage AI capabilities sooner and often on their own.
Since so much of my career has been in or around the OR, I decided to start there with use cases for conversational AI. Most examples I show are custom GPTs in ChatGPT that I created as prototypes. These could also be done in other applications like Poe or Claude AI. I’ll refer to these as bots, though I recognize that different companies use various names for essentially the same concept—something the OR is already very familiar with.
1. Navigating the Overload of Manuals—The Instructions for Use (IFUs) and equipment manuals in the OR are essential and vast. These documents must be readily available for reference and compliance with multiple regulatory agencies. However, locating the correct document when you need it most can be a challenge. Once located, they are also seldom light reads.
With conversational AI, instead of hunting down documents and relying on keyword searches, you could simply start a conversation using a bot that can access the manual’s content.
• How do I set up the Abar Laser?
• The Jackson Cautery won’t start.
• Teach me about the Kol irrigation system in two minutes
• How do I clean the GracD Fluid System?
• What are the safety precautions for the Claude Micro Cautery?
2. Tissue & Implants—Similar to Equipment Manuals, both IFUs, packaging inserts, and additional governing documents for tissue and implants could be interacted with via chat. AI could act as a conversational decision-support tool by answering specific questions related to guidelines like the AORN Association of periOperative Registered Nurses’ (AORN) Guidelines for Perioperative Practice, OR policies, or documentation from tracking systems. For instance, the nurse can ask, “How do I preserve cranial bone flaps?” The GPT would instantly extract and present the relevant information in an easy-to-read format.
3. OR Policy Creator—This bot could allow you to create policies based on standards and best practices from organizations such as AORN, the Association of Surgical Technologists (AST), and even The Joint Commission. I created my test bot to reference best practices, to not bring in content from outside of the best practice documents, to simplify the text to make it easy to understand, and to ask questions if clarification is needed. I also provided the hospital format so they can be easily copied and pasted into a document for edits and approval. This is also a good time to consider hospital policies on the use of AI.
4. Virtual Vendor—While vendors in the OR continue to be a bit controversial, there are many times when their knowledge is imperative. This one requires some collaboration. Creating an AI bot that references information or a knowledge base provided by the vendor could allow for AI conversations at 2:00 a.m. when the OR team is called in and using a tray or equipment they are unfamiliar with or when policy prevents vendors from being in the room.
5. Chatting With the Experts—While I mentioned AORN’s Guidelines for Perioperative Practice with Tissue & Implants, the potential is so much bigger. This is the reference for perioperative nursing practice. Bringing licensed expert content into a conversation could provide much-needed access and knowledge to overwhelmed new OR nurses and experienced ones who need to reference best practices. It could also be an ongoing companion to the essential AORN Periop101 curriculum.
6. Collaborative Cheat Sheets—There is a gold mine of siloed knowledge and information in the OR. While doctor preference cards provide some info on setup, case details, and intricacies, they aren’t always updated, and much of the seasoned staff’s knowledge remains in their heads. This takes us back to knowledge-based documents. Instead of siloed information in notebooks, phones, and personal memory, move this to a shared repository of knowledge that allows the new staff or those working in an unfamiliar service area to benefit from the decades of experience of their teammates. It may also prevent sifting through pages of notes on the preference card, which, if we are honest, we know staff sometimes don’t have the time or patience to read. Yes, a gatekeeper will probably still be required, but it is most likely a much easier process than doctor preference card note updates are today.
7. Global Unique Device Identification Database (GUDID)— The GUDID is a centralized database managed by the FDA that contains detailed information about medical devices, each identified by a Unique Device Identifier (UDI). I didn’t believe I would ever see a standardized device database in my career, but here it is. Unfortunately, many OR staff aren’t aware of its existence or the value that it can provide. It is also a massive database, so having AI do the heavy lifting of locating an item and its details could be a time saver. Connecting to the GUDID, conversational AI could allow the user to ask questions about the item’s type, whether it is a drain or implant, whether it contains latex, whether it is safe to use in MRI environments, and even provide vendor contact info.
8. AI Patient Support and Reassurance— In preparation for surgery, the patient is often told someone will call them a few days before surgery. This is already a stressful time for the patient, and it can be compounded by fear of the unknown and lack of control. A conversational AI bot can serve as a resource for patients preparing for surgery by providing clear and personalized guidance on the process. The bot may alleviate anxiety and help patients feel informed and reassured by answering questions about what to expect next. It could offer tailored information based on the specific surgical procedure, detailing pre-surgery steps, expected timelines, and day of surgery processes. By integrating this AI-driven support, healthcare providers could improve patient engagement and satisfaction, leading to a more positive overall experience.
9. OR Resource Image Viewer— There is so much to learn and remember in the OR. There are tens of thousands of instruments, equipment, disposables, and implants that you could be asked to locate and pull use immediately. An image viewer requires more than creating a local knowledge base, but the effort required could be well worth it. In this use case, the nurse or tech will ask the bot to show them a Deaver retractor or a specific catheter; the bot can then display an image with a description (huge bonus if it provides a location). Behind-the-scenes images need to be hosted, and keywords need to be mapped to these images in a database using a programming language such as Python. It would be even better if an API could access a system containing these images. This one is more complex and moving from being an AI Bot to being more of an AI Agent. Agents really make things interesting. I’ll save that for another day, as they deserve their own discussion.
The prototypes I’ve included were created as GPTs in ChatGPT but don’t discount other options, such as Claude AI.
10. Intelligent Visualizations and Dashboards— Claude AI allows you to upload data on things such as surgical case volume, turnover time, first case starts, or the content of your choice. You can then prompt Claude to analyze the data and suggest the best way to visualize it to communicate the content you’ve defined for the audience. You can also have Claude explain why certain methods are preferred, analyze the included data, and create a chart or even a dashboard! I went pretty basic with my content and a graph, but this took seconds, and I can continue a conversation with Claude, refining and reformatting the data as needed.
I’ll stop at ten, but know that these are just scratching the surface of what Conversational AI can do. I didn’t even get into the use cases for supply chain.
While I also didn’t go into the opportunities, integrating with AI tools like Microsoft Co-Pilot and Google Gems with Microsoft 365 and Google Workspace, the opportunities to connect email, calendars, documents, and create workflows make this list look like only the beginning.
Just think of what a focused team of experts at your organization could do to address pain points and processes with AI.
*While safeguards are becoming available, conversational AI is a relatively new technology. Though its potential is enormous, it requires careful and rigorous testing, policies, and prompt editing to help ensure it delivers the desired responses accurately and reliably.
Leave a Reply