Are you interested in ways that AI can improve sleep apnea surgery? So am I!

I just returned from a sleep apnea surgery course where there was a tremendous amount of discussion about how artificial intelligence (AI) revolutionizing medicine now and in the years to come. After decades of optimism (from others) and pessimism (from me), I agree we may be turning a corner in realizing potential, especially given the incredible pace of change in the use of generative AI outside medicine. As a child, I can recall listening to prophecies that computers would replace most of human work within a decade – over 4 decades ago – so I hope one can forgive my pessimism that was duly rewarded for years. Yet even a naysayer like me has come around to the tremendous potential that AI may offer for medical care, including in sleep apnea surgery.

To be clear, I am not an AI expect but understand the discussion about potential catastrophic implactions of AI wherever it is used. This post just discusses some of the most-apparent beneficial uses of AI in sleep surgery care, focusing on some that are currently possible or will be soon.

What is the low-hanging fruit for technology in sleep apnea surgery?

I love what I do, combining patient care with education and research in sleep apnea surgery. What I would love even more is to avoid time spent on repetitive, boring, and unproductive tasks and focus on the uniqueness of each sleep apnea surgery patient coming to see me. Mundane tasks include certain parts of medical care documentation and administrative work but also some of what I discuss with my sleep apnea surgery patients so many times a day: why we might consider sleep apnea surgery, the range of sleep apnea surgery options, and more. I believe this information is extremely important, but incorporating this into medical visits can take time away from discussion of other critical sleep apnea surgery information. More importantly, there is only a certain amount of information that a person can take in at once, so I believe that the general information I provide about sleep apnea and sleep apnea surgery can limit how my patient comes away from their office visit. When I was at USC, I developed a number of videos that I posted on my YouTube channel, including the one below:

While I do think it would be helpful for sleep apnea surgery patients to watch some of these videos before their appointment, I recognize that this does not happen often. My hope would be that sleep apnea surgery information could reach patients more simply, whether in the office or their personal device (if they choose). If patients received some of this foundational sleep apnea and sleep apnea surgery information in advance, they would have more time to digest it and get even more out of their sleep apnea surgery consultation. We could dive more deeply into questions that they might identify with more time to reflect on their special situation and goals as we consider sleep apnea surgery.

I currently take many steps to provide as much time as possible for my sleep apnea surgery patients to ask questions, both during and after their visits. One of the most important is our ability to send new patient forms electronically to sleep apnea surgery patients, for them to complete at some point before their appointment. Sleep apnea surgery patients sometimes wonder why our new patient forms are somewhat long (they take about 20 minutes to complete), but what this does is provide me with a greater sense of what we call the sleep history, saving about 15 minutes of time while providing higher-quality information.

Beyond my videos and website, I know that sleep apnea surgery patients are using these generative AI models. I am writing this post because the topic was featured at this recent course and because of a recent medical journal article. The April 2026 article in the Journal of Clinical Sleep Medicine examined the quality of sleep apnea educational information available in various sources: Google Search, general-purpose large language models, and a medically-specialized model (OpenEvidence). I will admit that I had not heard of OpenEvidence but was impressed that it seemed to offer the highest quality information (mean human reviewer score of 4.33 on a 5-point scale) over 30 questions related to sleep apnea (including sleep apnea surgery). I am a convert and will start using OpenEvidence to answer some of my own questions where I previously performed a Google search. I am excited to see how these tools develop going forward. Note: I have no financial or other interest in OpenEvidence.

What about some intermediate goals for AI in sleep apnea surgery?

Where to start? First, there are many wearable and so-called nearable devices that perform some sleep monitoring. Their use has tremendous potential as a long-term diagnostic (establishing the presence of sleep apnea) or outcomes (longer-term monitoring of sleep apnea surgery patients). Unfortunately, most of these devices utilize proprietary approaches to data analysis and have not been validated against gold standards. The long and short of it – and what I tell my sleep apnea surgery patients – is that the vast array of personal devices are not useless but are also not entirely useful. One achievable goal would be further development and careful evaluation of various technologies that enables these to go beyond what we currently have in sleep apnea testing: home or in-laboratory sleep studies that are generally done for 1-2 nights only. I know these devices are capable of so much more, but high-quality monitoring over time would be an advance in sleep apnea surgery on the most-basic level.

Second, so many of our evaluations can benefit from implementation of AI strategies, starting with interpretation (scoring) of sleep studies and sleep apnea surgery examinations like drug-induced sleep endoscopy. At this course, I learned how real advances are occurring related to AI scoring of sleep studies. One of our outstanding residents and I are collaborating with an engineering team here at UCLA to conduct our own research with AI and interpretation of drug-induced sleep endoscopy videos that are so important in sleep apnea surgery decision making.

More broadly, there is a mismatch between the tremendous number of patients with sleep disorders and the number of practitioners in sleep medicine. I believe we will soon see the most-straightforward patients proceeding directly to obtaining CPAP without needing a sleep study or, at a minimum, with fewer visits with sleep medicine providers. This will free up those sleep medicine providers to see the more-complicated patients that truly need their expertise. I expect there will be greater attention to monitoring sleep apnea patients, making sure we are doing everything we can to enhance CPAP comfort and adherence but also being proactive in identifying patients who would benefit from sleep apnea surgery evaluation and possible sleep apnea surgery.

There are endless opportunities for AI to streamline research in all fields, including sleep apnea surgery. These start with combining large datasets but also performing sophisticated analyses of small and large datasets. With the revolution in computing power available now at our fingertips, we should be able to accomplish so much more if we can just apply these tools properly.

How far can AI go in sleep apnea surgery?

I happened to be speaking about AI to a former professional baseball player, and we laughed that AI has not replaced athletes yet. However, he pointed out that the automated ball-strike challenge system has come to Major League Baseball, with generally-favorable reviews among the players (and we agreed the umpires will like it more over time). I do not think AI will replace humans performing sleep apnea surgery in my lifetime. However, AI can absolutely improve sleep apnea surgery care, whether improving the sleep apnea surgery itself or just patient care before and after sleep apnea surgery.

In my training, two pieces of advice I received from senior surgeons really stand out when I think about how AI can improve sleep apnea surgery. One was, “Surgery is just one detail after another.” The other was, “Surgery is designed so that everything is smooth even on my worst day.” AI seems perfectly suited to managing details, perhaps even better than a Type A person like me. Beyond that, AI should not have much variation between best and worst days. The point is: how we can best deploy AI to make surgery that much easier for patients and surgeons? There is so much potential for surgical planning and decision making.