Dr. Alpesh Patel (Host): Welcome to Better Edge. I'm Dr. Alpesh Patel. I'm the co-director of the Northwestern Medicine Center for Spine Health and a Professor of Orthopedic Surgery and Neurological Surgery here at Northwestern. Today, we're going to be revisiting a conversation we started back in 2024 with Northwestern Medicine spine surgeons Dr. Najib El Tecle, Assistant Professor of Neurological Surgery; and Dr. Srikanth Divi, Assistant Professor of Orthopedic Surgery and Neurosurgery, and we're going to talk specifically about the role of AI and robotic technologies in spine surgery into 2026.

Back in 2024, long, long time ago, many of these tools were still considered early, right? They were at the very early stages of clinical adoption, and we had as many questions as we had answers, I think, back then. So, we have the chance now to revisit this conversation because, since then, AI and robotics have moved from promise into our clinical practice.

Today's discussion is going to focus on what's changed, where these technologies have actually shown true clinical value, how we integrate these into our offices and operating rooms and in our care of patients, and emphasize, I think, a lot about how surgeon judgment continues to shape how we use these tools both now and in the future.

So with that, Dr. Divi and Dr. El Tecle, thank you for joining me and our listeners.

Dr. Najib El Tecle: Thank you for having us.

Dr. Srikanth Divi: Thank you.

Host: So, I think what we'll do is we'll pitch off. We'll start just by a quick recap because I think these conversations are always more effective when we all are using the same words and defining things with the same vernacular.

So just, Dr. Divi, AI, right? AI means a lot to different people. For the purpose of our discussion today, how should we think about AI?

Dr. Srikanth Divi: Yeah. It's a great question. I think, in the last four to six years, everyone has become accustomed to this term, because it's pervasive in our daily life. I think artificial intelligence, you can probably define this as something that replicates human intelligence, so a machine or computer that does that.

And, you know, one common example is like a self-driving car or any of the apps on the phone that give you recommendations. They use different tools, but that's where artificial intelligence comes in. You can think of it like as someone there behind the scenes kinda giving you recommendations.

And in the context of healthcare, and let's say specifically surgery, I would think of artificial intelligence as a tool, like either an application or a physical tool that helps a surgeon do their job.

Host: Got it. And when you talk about AI, it can mean artificial intelligence. I think a lot of us think about sort of self-learning computers from science fiction, but I think we're talking about something a little bit different, right? A little more tangible. You mentioned a tool.

Dr. Srikanth Divi: Right. I think, yeah, when people, you know, hear artificial intelligence, they may be thinking about robots that perform surgery on their own. Now, that may be something that happens in the future, distant future. But for now, I would think of artificial intelligence as something that helps a clinical team or a surgeon do their job effectively.

Host: Got it. Got it. And it certainly seems to be much more in the information flow and decision-making standpoint. And when we think about actually then, Najib, when we do the work that we need to do, talk about robotics. What does robotics mean to you as far as spine surgery is concerned?

Dr. Najib El Tecle: So, these two terms, people like to use them a lot and sometimes interchangeably, right? AI and robotics. But to me, when we're talking about robotics, it's more if AI sits at the software layer, if you want, at the data layer of the decision-making, robotics is what pertains to hardware, to us doing the work we do in the operating room.

In its current form, in no way is it something—as Sri mentioned—it's not a robot doing surgery by itself, but it's a tool that augments our technical skills in the operating room, allowing potentially for more precision in what we do. Yeah.

Host: No, I think that's a great summary of it, and I think we'll talk as we go along today, and maybe every now and then, let's compare to some of the robotics that are already been in the operating room for a lot longer, right, with other subspecialties. Our world in spine, and this is all still really new and innovative, and so it's exciting for us, but some occasional comparisons might be helpful.

So, I think that's a really good way to start in terms of definitions about AI being sort of information-driven, analytical data inputs and recommendations as outputs, if you will, and robotics as being some of the tools and the things we do in the operating room.

Let's talk about evolution. So, you guys were on this podcast a couple of years back, and maybe I'll start with Dr. El Tecle first. Since 2024, you know, what do you think has been the most meaningful way that robotics or AI has changed our daily clinical practice in spine surgery?

Dr. Najib El Tecle: I feel if we look at two years ago when we were having this conversation, at the time, there was a lot of hype, right? And then, people want to talk AI and then say AI this, AI that. And over the past two years, I feel things have matured a little bit where a lot of things went from being experimental. At the time, we were talking about automatic alignment parameters and things like that, and moved from being experimental to being operational where we use these in our clinics to kind of automate alignment parameters and properly plan for surgeries.

So, I think that move all happened at the same time as the field is maturing, as AI and spine surgery is maturing. And the hype in the beginning was, "Hey, this is going to be fast," and all that stuff, and we came to the realization, "No, this is not about speed." This is about precision. This is about decision-making. This is about taking all that data and seeing how it helps us guide the patient and do what we do in a better way.

Host: Yeah, I love that. Every new technology brings a lot of hype. And we've lived through these hype cycles, right, over and over again. And I love the example you shared about our automated measurements. And for those who don't know, our Northwestern Medicine Center for Spine Health is a collaborative effort between neurosurgery and orthopedic surgery, where we work together as colleagues and partners, along with our physiatrists and physical therapists.

And in there, we have a cutting-edge X-ray machine called the EOS machine that allows us to obtain amazing low-radiation images of patients that, to your point, gives us a ton of data to then apply towards analytics. And the one example you shared is this idea of automated measurements. So, we used to spend what felt like hours, but was probably minutes, but many minutes repeatedly measuring different angles and alignments and parameters in the spine, and now that's automated for us. And we see that continue to evolving. That's a great example.

Dr. Divi, how about yourself? What do you think has changed since 2024, in terms of either AI or robotics as far as our surgical care goes?

Dr. Srikanth Divi: Yeah, I think, at the last podcast, I remember us talking a lot about predictive analytics and how we're going to use machine learning, a specific subset of AI, to kind of help improve patient outcomes.

I think that's still in development, and that kind of goes hand in hand with the automated kind of radiographic parameter measurement. One example of a tool that we readily kind of adopt in our practice that I think has helped a lot, and you guys correct me if you guys have a different experience, but I think use of the ambient listener, Haiku. Just for those not familiar, this is a tool through Epic, which listens to our conversation with the patient and automatically generates a note. It's by no means perfect. But I think for me, it's largely changed my interaction with the patient and made it much more enjoyable. You have to guide it a little bit by, you know, telling it my specific interpretation of the imaging and the physical exam findings. But I think that's one AI tool that we have, in the last two years, that I've been using all the time.

Host: That's, again, just for everyone who doesn't have it, although it's filtered out, you know, pretty widely now, right? I read a number that about 40% of physicians in the US right now are using some type of ambient listening technology. But for those who haven't yet, the idea is to take a software that's listening to the conversation, applying large language models or natural language processing, and then translating that into a note. And that means that, Dr. El Tecle, you don't get to-- you don't have to type notes anymore. And you can actually look at a patient instead of looking at a monitor.

Dr. Najib El Tecle: Yeah. As you're saying, it definitely makes that time between you and the patient more valuable. You're focused on just talking to the patient, listening to what they have to say. And AI on the back end is making sure nothing's being forgotten, everything's going to transcribe a note. Yes, it's not perfect. But similarly, it's changed how I interact with patients and, in a way, made me more efficient. And I truly like that tool.

Host: So, that's a great—we had a couple examples, automated measurements for our X-rays, which we, again, as surgeons care a lot about. But broadly across healthcare is the idea of taking some of the workload of documentation off of our shoulders and allowing us to spend a little bit more face-to-face time with our patients. I think we like that. Our patients hopefully like that as well. Other than those two examples, can you share some other examples where you think we see the most impact from either an AI or a robotic technology? And for now, let's just lump them together. But Dr. Divi, what do you think of another example as we think about some of the complex spine surgeries that we take on at Northwestern?

Dr. Srikanth Divi: Yeah. I think a couple other things that the last couple years I've been using pretty routinely are patient-specific instrumentation, so that's patient-specific rods and interbodies. So for people that are not familiar, in spine surgery, as surgeons, what we view as critical is obtaining the optimal alignment for the patient, and we do a lot of planning before surgery with the angles that we mentioned and the desired outcome. But with a patient-specific rod, we plan all of this beforehand, and in surgery, we have the instrumentation matched to where we want the patient to be.

So, this kind of reduces the cognitive load intraoperatively in terms of stressing out about "Am I at the right alignment before or after? And these patient-specific interbody cages are also designed to help you get the alignment that you want and also achieve the nerve decompression that you want. And you can all do all this planning beforehand sitting down at a computer and be confident that you're getting your desired outcome.

Host: Yeah. It's that idea of using these technologies, especially around the AI and the informatics component to shape implants, right? You're talking about rods and interbody devices that we use to obtain fusions that match where patients need to be, right? And so, we've always forever and a day planned, but the planning was always one stage, the execution was another. And they were separate, right? And so, you were lucky enough to have, you know, many surgeons like us that have lots of years of experience to execute on that plan. But, you know, this maybe makes that more reliable. Is that the idea, to get it more consistent?

Talk a little bit, Dr. El-Tecle, if you will, as we introduce these new technologies into patient care. You know, I think we at Northwestern are extraordinarily thoughtful about how we bring these technologies in. We're not trying things out. We're not experimenting, right? Like, we're looking at things that are proven before we even introduce it to a patient. How do you approach and think about safeguards? How do we think about safety for patients when we are bringing in either AI or robotic technologies?

Dr. Najib El Tecle: I think like with any technology, any new technology in general, you need to master the technology before adopting it and using it on patients. And by mastering, I mean you need to know where it fails. Surgeons that know where technology fails are the safest surgeon at using these technologies. The same with AI, robotics, navigation, you name it. And it allows you to make the most out of that technology while making sure it's enhancing the care you provide to patients.

In a way, I do believe in these being a tool, but also a workflow. So, it's not a surgeon tool. It's a workflow for the entire operating room. Yeah. It's a workflow that goes from preoperative care to recovery, right? And through that process, I truly think mastering the pitfalls, knowing the ups and down, knowing where it shines, knowing also where it puts the patient at additional risk, that's critical to adopt these technologies safely.

Host: Yeah. I think one of the concerns that, you know, we have, we're also lucky enough at Northwestern to be educators, right? And we train neurosurgery residents, orthopedic surgery residents, and fellows that are doing spine surgery fellowships. So, I always think about the implication of these technologies on education. And I think what we worry about is that some of these technologies can be a crutch, right? They can somehow be a shortcut to the acquisition of critical skills, right?

So, I think what you mentioned just now about understanding what the limitations are, not putting, if you will, blind faith into any single AI model or any single robotic platform is, I think, really critical. I think we do a good job of emphasizing that here. But I think as we know, it's been one of the struggles, right? Because you want to get to automation, and you want to get to consistency, but yet you need to teach.

So, Dr. Divi, maybe can you talk to us just a little bit about education and how do we not just rationalize, but maybe integrate these technologies into how we train, you know, surgeons of the future?

Dr. Srikanth Divi: Yeah. That that's an excellent question. I think I view this, you know, in a couple different lights. One, I think it actually augment education. For example, like a junior resident who is just starting out, spine anatomy can be very complex, especially in like a complex deformity type patient.

And having, you know, a robotic navigation system or any sort of navigation system and applying kind of AI-based planning tools for trajectories for screws can help them understand where the screw is supposed to go if there's a narrow corridor, and understand the pitfalls of whether, you know, they're erring laterally or inferiorly. And it can help them visualize.

As they get older and progress through their training until they get to the level of a fellow, I think we then may have to stress the importance of doing that on your own, doing that mental exercise on your own, and emphasizing to that trainee, try to do this with your intuition and your landmarks and your training, and then only use the technology as a safeguard or a double check, basically.

That's kind of how I've been thinking about it. And as we move forward, you know, we can help create surgeons to use this technology to educate them and keep it as a safeguard without them totally being reliant on it.

Host: Yeah. No, definitely. You know, again, my experience on this, and I'm wondering what you guys think, is it took, as you mentioned, a really complex anatomy, right? With lots of structures at risk between vascular structures, neurological structures at risk in how we use different implants in and around the spine. It took that complex three-dimensional anatomy that we spend years and years and years developing in our head, and it kind of lays it out on a platter for you, right? So, it makes the barrier to entry much lower, which is great, because we don't want necessarily everyone to go through a steep learning curve. But at the same point, how do we still learn?

I think one of the things you mentioned is sort of fitting the technology, if you will, to the level of education and training. I'm curious, Dr. El Tecle, what you think, you know, as you work with the neurosurgery residents and our neurosurgery spine fellows about these technologies?

Dr. Najib El Tecle: It's pretty much a very similar experience, where in the very beginning, we're relying on that technology. "Hey, get to know the technology, get to know the anatomy. Let's have the technology help you further understand the anatomy in a faster way." And that progresses as they mature in their residencies and their fellowship. But by extension to that same idea, that's where the technology, in a way, and I think we talked about this last time, democratizes spine surgery, where now spine surgery is more accessible in the sense that by laying that anatomy in front of you, it's possible for these surgeries and some surgeries that in the past used to be exclusive to places like Northwestern now to be happening at, say, smaller hospitals, because the technology has made that available.

Host: Yeah. I think that's a great thing, which is we want more access to care. We know that's a big barrier. And at the same point, I think one of the obligations that we've taken on at Northwestern and our Center for Spine Health is to make sure we're training that next group of surgeons to be able to use these technologies, but then also to use the technology in their head that they were born with, right? Their brain and their cognitive capabilities to their maximum amount.

So, this then kind of leads to a next conversation that we hear a lot as we as surgeons talk sort of off camera and off microphone, which is AI and robotics standard of care. And it's a dicey topic to talk about. We'll talk about in the sense of do you think that AI and robotic-assisted techniques or technologies are starting to influence what we think of as standard of care? Or do we think that there's still enough flexibility out there that surgeons, you know, in a good conversation with their patients can, you know, use their judgment and adopt technologies as they see fit?

And then, let me ask you current state, for Dr.Divi, and then Dr. El Tekle, in 10 years, what do you think it's going to be? So currently, do you think that AI and robotics is standard of care, like you have to use it or you're not providing the best care, or is it still an option?

Dr. Srikanth Divi: No, I think at this point, it's not standard of care because it's not as widely adopted. There's a lot of reasons for that. There's barriers. Cost is a huge one. And, you know, we have to prove that there's that marginal increase in safety and patient outcomes, and that is worth the cost. So as for now, I think we're working on it. I think we see the perceived benefits, but it would be hard to say it's standard of care right now.

Host: All right. Dr. El Tecle, what do you think? Put your future hat on. Ten years from now, is it going to be standard of care that everyone out there performing spine surgery needs to use AI and robotics?

Dr. Najib El Tecle: There's the saying that people tend to underestimate the progress that happens in two years and overestimate the progress that is the opposite way. Underestimate the progress that happens in 10 years and overestimate the progress that happens in two years, right? And I don't want to fall prey to that.

Host: So, you know the trap, you know that heuristics. Yeah. So, what are you going to say? That's the trap.

Dr. Najib El Tecle: I think standard of care in a way is like a lot to ask of AI. I definitely see AI being like a surgeon co-pilot, right? Will it be the pilot, which would be standard of care? I don't think we're going to be there. I think probably it's going to be giving us more and more information. Robotics is going to make things safer. AI is going to allow us to go through that journey of care from pre-op to recovery and do better for the patient's standard of care. That's where I'm going to say. For now, I don't think so.

Host: Yeah. I think standard of care might be a loaded way to say it. I think it may be more pervasive, and it may be more commonplace. And I think that my take on this is whether you talk about two years or 10 years, and I'll go over and under on my estimation, that it's going to be more and more common for surgeons to utilize AI for sure, because that's informatics and decision-making, and maybe more common, in robotics, depending upon how the robots evolve over time, right?

And again, we're lucky enough at Northwestern to do a lot of the research that fuels this innovation. So, maybe we'll pivot for a couple of minutes to talk about that, you know, at Northwestern and how lucky we are to have some of the infrastructure around us. We have these technologies in current state, as we care for patients today in 2026.

But as we think about future state, let's talk a little about research. And I'll start with you, Dr. El Tecle. Tell us just a little bit about what you and your team are working on from a research standpoint around, let's say, AI or machine learning and robotics that might influence us down the road.

Dr. Najib El Tecle: Absolutely. We've looked at several, if you want, topics and themes that we've covered with AI, some of them pertaining to patient education, appropriate education material for patients. How do you drive whatever is currently online, make it accessible to patients? But the area I'm most excited about is this idea of objective outcome measures where, as you know, for a while we've relied on subjective assessment of outcomes in spine surgery. And now, we're looking at collecting more detailed data, things like step counts, and even now that we have more granular data, heart rate variability and things like that to measure how patients do before and after surgery. That's part where AI can augment that analysis and give us more insight into how these patients are doing before and after surgery. That's the part I'm most excited about.

There's definitely a lot more in terms of surgical planning. There's the part about predictive analytics that Sri mentioned earlier. And while we talk a lot about planning for surgery properly, how do we verify that we're executing properly, and how do we predict what things are going to look like after surgery. And in a way, there has been this idea of precision medicine, right? That's been circulating forever. Now, I think AI augments precision medicine and makes it truly individualized care for the spine patients.

Host: I love the idea of objective measures, right? Because our patients often want to know, "Am I going to be able to do this better or that better?" And when we talk to them in abstract terms, you know, around subjective outcome measures or other things we might utilize in research, it might get lost in translation a little bit.

So, I love the idea of having targets and goals for patients and the wearables that you mentioned around step counts is super cool. How about Dr. Divi? Tell us a little bit about what are you thinking in terms of the research that you're helping us lead at the Center for Spine Health around AI and ML work. What are you excited for?

Dr. Srikanth Divi: So, last time we talked a lot about kind of our shared research of machine learning and outcomes, and I think in the last couple of years, we've kind of tried to keep pace with the emerging AI technology. It's always hard, but embracing large language models to help with decision support.

And I think that's going to be an important tool for spine surgery. It's already being used in a lot of other medical fields. But I think specifically with spine, how it may really help is because decision-making is very complex. You can have the same kind of condition, injury, among, like, let's say three different patients. And based on their health status, their functionality, you may come up with three different options: non-operative, operate later, or operate now.

And this is where I think a decision support tool can help us filter by taking in all the information that's specific to that patient and telling us, "Hey, this patient actually might benefit from surgery now or never." And just like you might cause more problems by doing surgery. And that's where I think doing this large language model research, which we're conducting on some of our retrospective data that we have, can help.

I'm also excited to kind of maybe take the next steps, is integrating that decision-making with the vast repository of like imaging data we have. And if we combine that with patient-specific outcome measures with, you know, the subjective measures that we have, along with maybe objective measures, then we can correlate lots of things, the medical history, their exam, along with the imaging objectively shows, not just like our interpretation or radiologist interpretation of it, and then what the expected outcomes are. Because I think if you combine all that, then you can tell a patient, "Here's what realistically can get better, and here's what probably won't get better."

Host: But specifically for them, right?

Dr. Srikanth Divi: For them, exactly.

Host: I think a lot of patients, you know, and a lot of us, as you get older, you become a patient. What you appreciate is the knowledge and insight and data. But you'd like to know again, okay, I get it, across the thousand patients, this is what it looks like. But I think most people want to know, "What does this look like for me?"

Dr. Srikanth Divi: Exactly.

Host: And do you think that we'll have some of the tools available in the coming few years to be that personalized?

Dr. Srikanth Divi: I think so. I think, if we have the right tools to like train the model that we want, I think we can get to a level of accuracy, you know, where we could tell a patient, you know, this is 80% or 90% effective for this problem that you have. I hope we can get there. But I think we still need to build that out.

Host: Yeah, for sure. And as you mentioned earlier, we can't put blind faith in these platforms, right? Like, we need to leverage them back into the reality that our patients are living in. But it is exciting.

One last thing as we wrap up here, if there's a practical takeaway, so if someone listening to our conversation right now, let's say they're a surgeon or even a patient thinking about surgery, what's a practical takeaway you give them today? I'll start with you, Dr. El Tacle. Like, if they're thinking about or curious about AI or robotics as part of their care plan, what's a practical step they should take now to understand it better?

Dr. Najib El Tecle: I think like adopting any new technology, getting to know the technology really well becomes very important. As a surgeon, there's definitely plenty of resources, right, to talk to other people who are using the technology and getting to know the technology better before using it. And as a patient also, I think, there are two layers. One, the part that you do by yourself, like we all probably are going to ask ChatGPT for any symptoms we have or whatever, but also there's the part—

Host: Or any other model.

Dr. Najib El Tecle: Or any other model. Yeah.

Host: We're not tied to Chat, but yeah.

Dr. Najib El Tecle: But also, there's the part of establishing that human relationship and connection with a surgeon who's going to take good care of you, where I think we're not past that point. I truly think that still is number one. That's the physician-patient relationship. So, talking to people that do these kind of things.

But as we discussed earlier, this is not necessarily the right technology for everyone, and knowing who are the right patients for it becomes critical and crucial.

Host: Yeah, it's definitely not one-size-fits-all. There definitely can't be a you always do this or always do that. And I love what you said because I think that's why we get along so well in our group at Northwestern, that we have such a good relationship between, you know, orthopedic surgery and neurosurgery, is that we all have that same feeling, which is all the technologies in the world, all the tools in the world, in the end, it's still like that conversation with a person across the table from you or across the office from you, and that's what matters to us.

Hopefully, some of these tools make it easier. You mentioned patient education. We talked about less documentation, less sort of typing at computers. Those are all good things, I think, and maybe it starts to inform better decisions. How about Dr. Divi, from your view? If we were to say again, practical takeaway, current state, first maybe as a surgeon, if you're talking to another surgeon, how do I implement this? And then, pause and tell us, you know, as a patient listening, how should I think about these technologies right now?

Dr. Srikanth Divi: Sure. As a surgeon, looking to adopt some of these newer technologies, I would tell them There are some really helpful tools here. And what they can do is take, you know, the stress level of daily practice, decrease it, and make it enjoyable while not affecting kind of patient outcomes or maybe even improving them. But what that takes is some level of vigilance and not just blindly relying on it, but incorporating it slowly into practice and making sure it tracks with how you would plan it. So, double-checking everything it does, and then relying on it more.

You know, for example, one, you know, the notes that we get from this generated LLM, we still have to go back and edit it. there are still plenty of mistakes. Just like that with our levels for planning and patient-specific instrumentation, you still have to double-check everything. But as you become more confident with it, it's a tool that I think decreases your daily stress level and makes it enjoyable to kind of treat patients.

From the patient perspective, I think it's actually opened up big avenue for them to get educated on surgery. And as Dr. El Tecle mentioned, you know, patient education is huge. I think we can start to develop tools that, you know, create guidance. And I think this is already being done, for example, in joint replacement surgery, where you have like a recovery aid app that tells them at every time point what they should be doing, et cetera. I think we can really apply this to spine patients for sure.

It should hopefully decrease the number of calls and all the constant, you know, questions that we get. But, you know, I'm also noticing patients in the office asking more pertinent questions about, you know, certain terminology or, you know, certain things that they see on their imaging, and they'll feed their notes to the large language model and ask them about its interpretation.

But as a surgeon, we also have to do that, so we can be ready to answer their questions. Be like, "Actually, the model says this, but here's what it's trying to say." And we all know the limitations of hallucinations in the models, right? So, it's not always accurate, and you have to be ready to be the expert and, like, call it out.

Host: I think that last part's really important, right? We can't ignore these technologies, because there's some element of that is sort of saying, "Well, you know, let's wait till they're proven, and we'll just ignore it." It's in the hands of our patients. It's in the hands of our partners and our colleagues and our residents and our trainees.

We have to understand it, and to your point, we got to be experts on it, right? Like, this is what the task we've taken on as a group of surgeons here at Northwestern Center for Spine Health to say, "Listen, we're going to become the experts, so we can give best guidance."

Well, listen, that was awesome. It went really fast. I want to thank you guys both for your time. Thank you, Dr. Divi, thank you, Dr. El Tecle, for sharing your insights and experience. And again, thank you to the listeners for spending the last half hour or so with us. So for more expert insights and clinical updates from Northwestern Medicine, visit breakthroughsforphysicians.nm.org. And thank you so much for listening and tuning in.