Episode #151 Artificial Intelligence and Anesthesia Patient Safety

Hello and welcome back to the Anesthesia Patient Safety Podcast.  My name is Alli Bechtel, and I am your host. Thank you for joining us for another show. To kick things off today, I asked Chat GPT to help me with the intro to the show. I asked, “Can you write an introduction to a podcast episode about anesthesia, patient safety, and artificial intelligence?” This is what Chat GPT came up with:

“Welcome to this episode of our podcast where we explore the fascinating intersection of anesthesia, patient safety, and artificial intelligence. Anesthesia is a critical component of modern medicine, allowing for painless surgery and other medical procedures. However, it also comes with potential risks, including adverse reactions to medication and errors in dosing. This is where artificial intelligence (AI) comes in, offering the potential to enhance patient safety by optimizing anesthesia dosing and predicting adverse events. In this episode, we will dive into the latest developments in AI-powered anesthesia and its impact on patient safety, as well as examine the potential benefits and challenges of implementing this technology in clinical practice. Join us as we explore this cutting-edge field and its potential to revolutionize the way we deliver anesthesia and care for our patients.”

Before we dive into the episode today, we’d like to recognize GE Healthcare, a major corporate supporter of APSF. GE Healthcare has generously provided unrestricted support to further our vision that “no one shall be harmed by anesthesia care”. Thank you, GE Healthcare – we wouldn’t be able to do all that we do without you!”

Thank you so much to Chat GPT for helping to kick off the show today. Our featured article today is from the February 2023 APSF  Newsletter. It is “Artificial Intelligence, Patient Safety, and Achieving the Quintuple Aim in Anesthesiology” by Jonathan Tan and Maxime Cannesson. To follow along with us, head over to APSF.org and click on the Newsletter Heading. First one down is the current issue and this article is right at the top of the page or you can click on the fifth one down, Newsletter archives, then click on February 2023 and you will see the article at the top of the page. I will include a link in the show notes as well.

And now, let’s get into the article which will introduce AI as an emerging technology and describe a practical framework for anesthesia professionals to understand the impact of AI on perioperative patient safety. Over the past 20 years, there have been significant technological advancements and exponential growth of data and computer power leading to the development of artificial intelligence, or AI, tools that can be used in the perioperative setting. We are at the brink of being able to apply these tools to further improve patient safety and quality of care going forward. AI has the potential to improve patient safety by allowing the clinician to gather large amount of data from different sources and effectively assist the clinician to synthesize the data and then to make better,  more informed medical decisions within a complex health care system. Another consideration for AI to improve patient safety is by integrating into the workflow of perioperative patient safety and quality leaders, patient safety scientists, and health care system leaders. AI may be used to improve patient safety by assisting with policy decisions that can help identify, assess, and mitigate threats to patient safety and do so on a large scale.

Let’s talk more about patient safety and emerging technologies. The authors provide this definition for patient safety: “the absence of preventable harm to a patient and minimizing the risk of harm in health care delivery.” There has been considerable improvement in anesthesia patient safety over the past thirty years with anesthesia professionals leading the way in research, quality improvement, adoption of technology, and incorporation of engineering principles to address threats to anesthesia patient safety. Going forward, continuing to provide safe anesthesia care is a challenge due to the following:

• Complex patients with multiple significant co-morbidities
• The speed of care delivery
• Scale of health systems
• Challenges in multispecialty communication
• Massive volume of data generated

There is a call to action for anesthesia care teams to increase their knowledge, presence, and effectiveness across the perioperative and health care environment to continue to work towards improved patient safety. AI may be able to help, but before we roll out the red carpet for AI, anesthesia professionals need to understand the principles of AI, the possibilities, the risks, the ethics, and the use of AI in clinical practice. Anesthesia professionals will need to collaborate with data scientists, computer scientists, data analysts, and artificial intelligence experts for this emerging technology to be successful in improving anesthesia patient safety.

So, how can artificial intelligence be applied in anesthesiology? Before we get to answer this question, let’s take a step back and first ask, “What is AI?” The authors define AI as the ability of a computer or device to analyze a large volume of complex health care data, reveal knowledge, identify risks and opportunities, and support improved decision-making.” Here are some of the major technologies used in health care: machine learning, natural language processing, and combining AI with clinical decision support with graphic user interfaces.

First up, let’s talk about machine learning which is a statistical technique for fitting models to data with the computer “learning” how to understand the data after first using training datasets. This technology can be scaled up to more sophisticated forms such as neural networks and deep learning. Have you used machine learning in anesthesiology? Available technologies include studying which variables could be used to predict postinduction hypotension from the electronic health record data, determining the bispectral index or BIS value based on data from the infusion history of propofol and remifentanil, or predicting postoperative in-hospital mortality using preoperative and intraoperative data.

The next technology is natural language processing which is a form of AI that is able to extract relevant information from unstructured text data. Okay, we definitely need an example of this type of AI. Natural language processing may be used in a retrospective study to evaluate unstructured free text, such as medical conditions in the electronic medical records, and then using AI, generate an automated preanesthetic evaluation report. Wow, that is so cool. This study was recently completed and published in Anesthesia and Analgesia by Suh and colleagues. I will include the citation in the show notes as well. The study evaluated how often the natural language processing software was able to recognize medical conditions compared to anesthesia professionals. Results included the following:

• Natural language processing picked up relevant conditions that were missed by clinicians in 16.57% of the cases
• Relevant conditions were missed by the AI in only 2.19% of the cases.

This type of AI may have significant implications to enhance the ability of the anesthesia professional especially in a busy perioperative environment with limited staffing resources and may go a long want to improve patient safety and the quality of care.

Another example of natural language processing is in clinical decision support systems which may be part of the anesthesia information management system in the form of electronic reminders to the anesthesia team in the operating room about perioperative antibiotic dosing, use of postoperative nausea and vomiting prophylaxis in high-risk patients, and blood glucose management. AI is already being used to improve anesthesia patient safety. A recent meta-analysis demonstrated improved compliance with perioperative antibiotic prophylaxis with clinical decision assistance. Check out the article by Simpao and colleagues, “A systematic review of near real-time and point of care clinical decision support in anesthesia information management systems.” I will include the citation in the show notes as well. Going forward, we may see an expanding role of AI in clinical decision support in providing recommendations on the ideal antibiotic given the patient’s electronic medical record information, medical history, and surgical procedure or for earlier detection of clinical deterioration and provide real-time clinical decision support for optimal management of intraoperative physiologic changes.

Have you heard of the Triple Aim? Or the Quadruple Aim? Or maybe even the Quintuple Aim? These correspond to health care delivery goals. Check out Figure one in the article and we are going to review it now. It all started with the Triple Aim which was described in 2008 and includes the following:

1. Patient Experience
2. Population Health
3. Lower Costs

This was expanded in 2014 to the Quadruple Aim with #4: Clinician Well-being. In 2022, the fifth aim was added to complete the Quintuple Aim with Health Equity. We can use the quintuple aim as a lens to understand the direct impact of AI on perioperative patient safety.

Let’s look a little closer at the Quintuple Aim, which is the proposed next step in improving patient safety and quality of care delivered. The Institute for Healthcare Improvement introduced the Triple Aim in 2008 which kicked off the conversation about health care transformation with the three keys of patient experience, population health, and lower costs. Clinician well-being was added in 2014 after research revealed that decreased clinician engagement and increased burnout was a threat to patient safety and quality health care. These aims were recognized by the National Committee for Quality Assurance and the Joint Commission. The fifth aim, advancing health equity, was proposed in 2022 to highlight that in order to achieve high-quality and safe patient care for populations and the other four aims there must be a focus on actively measuring, studying, and addressing disparities.

Next week, we are going to return to our conversation on the relationship between anesthesia patient safety and artificial intelligence by looking at it through the lens of the quintuple aim. There will be a quiz, so make sure you review the 5 aims. To close out the show, I asked Chat GPT to come up with a closing, so I will let the AI take it away:

“Thank you for tuning in to this episode on the fascinating topic of artificial intelligence and anesthesia patient safety. We hope you found the discussion insightful and informative.

As we’ve learned, AI technology has the potential to significantly improve patient outcomes and reduce risks associated with anesthesia. By analyzing vast amounts of data and providing real-time feedback to clinicians, AI can help optimize dosing, predict adverse events, and even detect subtle changes in a patient’s vital signs before they become critical.

However, as with any new technology, there are also challenges and ethical considerations to be addressed. It’s important that we continue to carefully evaluate the role of AI in anesthesia and ensure that patient safety remains the top priority.

We encourage you to continue exploring this exciting and rapidly evolving field, and to stay informed about the latest advancements and best practices. Thank you for joining us, and we look forward to bringing you more thought-provoking discussions in the future.”

If you have any questions or comments from today’s show, please email us at [email protected]. Please keep in mind that the information in this show is provided for informational purposes only and does not constitute medical or legal advice. We hope that you will visit APSF.org for detailed information and check out the show notes for links to all the topics we discussed today.

Until next time, stay vigilant so that no one shall be harmed by anesthesia care.

© 2023, The Anesthesia Patient Safety Foundation