Decision and dexterity: Engineering the next era of surgical robotics with AI

To understand the impact of digital surgery, Surgical Robotics Technology sat down with Andrew Savarese, Global Head of Digital Surgery at PA Consulting, a global innovation and transformation consultancy, at the firm’s Boston innovation studio. Drawing on his projects with the world’s leading surgical robotics companies, Andrew offers a front-line view as to where the sector is headed — and how to get ahead.

Improving surgical outcomes ultimately comes down to enhancing both decision-making and dexterity. What does it take to meaningfully advance those two capabilities — and how are we getting there?

After almost 20 years of being in operating rooms, launching robotic platforms, a key cause of surgical errors is often the correct decision that wasn’t translated into a dexterous movement. From day one, the benefits of robotic systems have been that we can miniaturise, scale, and filter human movement, enhancing precision and surgical accuracy and making the procedure reproducible.

All this hinges on the connection between the surgeon and the robotic platform. If user inputs, whether digital or physical, aren’t designed to be a natural extension of the surgeon, no amount of engineering complexity will translate the surgeon’s intention into accurate movements. The system must be designed to focus on what the surgeon is trying to accomplish. When done artfully, as I saw first-hand with Intuitive Surgical or Mazor Robotics, the robot is truly an extension of the surgeon’s own hands.

One of the things we spend a lot of time on with clients is translating surgeons’ needs into clear requirements that define what we’ll build and why. We harness core engineering principles to ensure that, whether it’s a single or multi-port robotic system for soft tissue surgery or an endoluminal system navigating the vasculature, it has the dexterity to execute every task, regardless of the complexity. Across the 25+ robotic platforms we’ve worked on, we blend human factors, design, and engineering expertise with our clients’ clinical domain expertise to develop platforms that change the lives of patients.

Back in 2001, the first DaVinci cases turned surgical movement into 1s and 0s. Today, we finally have the advanced algorithms that can process it all. When this data is funnelled through a digital platform, the algorithms within can create an insight-driven surgical experience, guiding the procedure like a safe pair of hands.

But the art we are placing into our developments now is capturing years of training, clinical papers, case observations, and surgical experiences to determine when and why decisions are made. Now, split-second pivots intra-operatively lead to small tweaks in a workflow, all to drive a better patient outcome. At present, algorithms don’t yet understand those random course corrections, but in the near future, they will. When we reach that point, however, we can’t create analysis paralysis for surgeons. Similar to the robot as an extension of the human hand, these platforms must be extensions of the clinical brain, delivering the right insight at the right time to seamlessly augment the human.

One question we get all the time from clients is how to realise this digitally-informed future – how they can equip systems with the ability to capture data from arms, instruments, and video feeds, with built-in telemetry to determine system health.

Data is like fruit; it’s most nutritious when ripe. But, just like an apple that’s past its prime, if I complete a procedure today and don’t analyse the data for a year, it won’t be as crunchy. As such, architect your system for extensibility and build in the hardware that meets today’s needs while also allowing you to unlock tomorrow’s capabilities just by flipping a switch.

Of course, the sheer volume of data generated by robotic systems is immense, and without robust computing capabilities, there’s no way to turn terabytes into insights. Fortunately, we’ve reached the point where computational power can find smaller form factors at lower integration costs. Edge AI can tell me how the force I apply to an end effector during colorectal surgery differs from a thoracic or urologic procedure, thus altering the way I use the robot. Understanding the specifics of each procedure, as well as the robot’s performance, is vital. Blanket insights will do more harm than good, especially when the ultimate goal is for the robot to enhance the surgeon’s ability while simultaneously lowering the cognitive burden for the entire surgical team.

AI feels like the new “must-have” in surgery, much like robots were a decade ago. But is it truly essential, or are we chasing hype?

Well, if we’re serious about improving surgery and ultimately democratising access to it, we have to limit the variability of outcomes. That’s where AI, or more accurately, augmented intelligence, becomes essential. The amount of data generated in an operating room in a split second is massive. You’ve got telemetry data, radiographic images, anaesthesia readings, data from the robotic system, tissue response — especially in laparoscopic or navigated procedures — all coming in at once. Whether it’s a solo surgeon or a full team, it’s simply too much for any human to process in real-time.

The best surgeons aren’t consciously analysing everything. They operate on instinct, experience, and thousands of hours in the OR. They know how tissue should feel, they recognise when something’s off, and they react instantly — often without being able to explain how they knew what to do. It’s like a sixth sense. Now, if we want to scale that level of precision across a surgical speciality, we need platforms connecting data across the globe and AI to process information in real time. It’s not about replacing the surgeon — it’s about augmenting them, supporting them with insights they wouldn’t otherwise have.

And here’s the bigger opportunity: today, if you or I need surgery, our outcome depends heavily on the individual surgeon, where they trained, how they’re feeling that day, and how they’ve done similar cases in the past. But what if we could pull data from a global dataset of surgical outcomes while also understanding the patient’s health history, baseline quality of life, and their desired outcome to generate the best insights for their care.

That’s where AI really shines. It enables more personalised, data-driven decisions, not just based on the condition, but on the person. That’s the level of care we’re moving toward. And honestly, it’s not possible without AI.

So, what’s the smart way forward? And if you had the power to change just one thing to keep us on the right path, what would it be?

We can’t treat AI as a shortcut. Just like with surgical hardware, we need rigorous product development, with defined requirements and proper validation. We can’t skip those steps. And we can’t overlook the data. AI is only as good as what we feed it. Right now, data quality and consistency are major roadblocks. Something as simple as the time of a procedure can be recorded differently across hospitals. That inconsistency makes it harder to train reliable systems, which is a huge challenge. In a recent project to develop a clinical registry for the American College of Emergency Physicians, we found 47 different ways that healthcare professionals enter a date into their systems. That’s just one example of the complexities of creating clean, consistent data when developing these platforms. And AI shouldn’t be making calls like “this is cancer.” It should say, “Hey, something looks off here — take another look.” It’s a tool to enhance expertise, not a substitute for it.

If I could change one thing, it would be unlocking access to data. Across the industry events I attend, I constantly hear that companies have the platforms, the algorithms, the vision, but can’t move forward without the right data. And we still don’t have alignment across the ecosystem. Who owns the data? Who can share it? How is it used? Until we fix that, we’ll end up with brilliant software and no fuel to power it.

There’s incredible potential out there — platforms that could transform how we approach entire procedures. But without access to high-quality, shareable data, that potential stays locked.

This article was first featured on Surgical Robotics Technology.