Lung cancer is the deadliest cancer in the world, and the numbers have not moved as fast as anyone would want. Screening is getting better. But finding a nodule is only the beginning. Many are tiny, sitting in parts of the lung that standard tools simply cannot reach, and patients can spend months in a holding pattern of scans and inconclusive results while the clock keeps moving.

Joanna Nathan started Prana Surgical with two-cofounders because that holding pattern is not inevitable. She is building a surgical tool that lets physicians localize and remove small pulmonary nodules without resorting to open surgery or disproportionate tissue removal. This past November, Prana Surgical ran its first early feasibility clinical trials in Melbourne, Australia. Some of those nodules were 4 millimeters. Peripheral. The kind that conventional biopsy tools often struggle to reach.

Her thinking on AI is direct, and it carries weight because of where she has sat. Joanna has been a research engineer, a product developer, a venture investor at Mercury Fund, and a new ventures manager at Johnson & Johnson’s medtech innovation center. She has evaluated companies from the investor side and built from the founder side. That combination makes her hard to dismiss.

The team at Prana Surgical uses AI every day. The device does not. That is not an oversight. It is a deliberate line, drawn by someone who understands both the technology and what happens when medical devices fail. In this conversation, Joanna explains where that line comes from, why she believes parts of the industry are moving faster than the evidence warrants, and what it will actually take before AI can be trusted closer to the point of care.

Foundation: Understanding Joanna and Prana Surgical

1. You have built a career that spans research, product development, venture investing, and founding a medical device company. How do you introduce yourself, and what connects all of those chapters?

Most of my career has been spent trying to move ideas from the research stage into something that actually helps patients. I started out as a biomedical engineer working in translational research, which is where you first see how exciting turning an idea into an early prototype or initial pre-clinical testing can be. Later I moved into product development, where you learn very quickly that building something reliable enough for medicine is a completely different challenge. I also spent time on the investing side and at Johnson & Johnson’s Center for Device Innovation, which gave me a window into how companies get built. Founding Prana Surgical pulled all of those experiences together. It’s the same problem I’ve always been interested in, just from the inside this time.

2. For readers who aren’t familiar with Prana Surgical, what does the company do and what problem are you solving?

Prana Surgical is building a minimally invasive tool called the Prana System, designed to remove very small lung nodules. These nodules are often picked up on CT scans, especially now that lung cancer screening is becoming more common. The challenge is that many of them are only a few millimeters in size and can be difficult to reach reliably with existing tools.

That leaves patients in a frustrating spot. Doctors can see something suspicious, but getting a clear answer isn’t always straightforward. The Prana System is designed to let physicians localize and excise those nodules directly while preserving as much healthy lung tissue as possible. Instead of taking a tiny biopsy sample or removing a large portion of lung, the goal is to remove the nodule itself.

3. You recently completed its first clinical trials with the Prana System for small pulmonary nodules. What did this milestone prove to you about the clinical need, and what questions does it raise for the next phase of validation?

The early cases reinforced how real this problem is for clinicians. Some of the nodules we treated were extremely small and in parts of the lung that are hard to reach with conventional tools. Seeing physicians able to localize and remove them confirmed that there is a meaningful gap between detecting these nodules and having a reliable way to act on that information. At the same time, early feasibility studies are just the beginning. The next phase is about expanding the number of patients and understanding how the system performs across different clinical settings. Medicine moves deliberately, and rightly so. The focus now is on building the evidence needed to show that our approach can work consistently.

4. For a patient with a small nodule, today’s path can involve repeat scans, uncertainty, and hard decisions about biopsy or surgery. In practical terms, what are you trying to change about that journey? Where do you want the experience to be faster, safer, or more precise?

Right now, detection often comes long before a clear plan. A patient might learn they have a small pulmonary nodule, but the next step is often watchful waiting. That can mean repeat scans every few months and a lot of uncertainty in between. In some cases a biopsy is attempted, but very small nodules can be difficult to sample accurately. What we’re trying to do is close that gap between finding something and being able to deal with it. If physicians have a way to access and remove these nodules earlier, it could shorten that long period of uncertainty. The hope is that patients can get answers sooner, and in many cases the disease can be addressed before it progresses.

The Role of AI for Prana Surgical today

5. Let’s talk AI. Your team uses AI across product development, clinical work, and regulatory support, but the Prana System itself does not use AI. Why keep AI out of the device, while leaning on it inside the company?

Inside the company, AI is incredibly useful as a working tool. It helps us organize large amounts of information, draft documents, and catch inconsistencies in things like regulatory files. In a small company where everyone is wearing multiple hats, that kind of support can make a real difference.

Putting AI directly into a medical device is a different conversation. When a tool is being used on a patient, every part of that system has to behave in ways that are predictable and well understood. AI models can be powerful, but they can also behave in ways that are difficult to fully characterize. Until there is stronger evidence and clearer regulatory pathways, we are comfortable using AI to support the people building the technology, but not yet to guide the device itself. 

6. Take us inside Prana Surgical’s daily workflow. Where does AI show up most often, and which teams lean on it the hardest?

It shows up mostly in the background work that goes into building a medical device company. Our product development and quality teams deal with huge amounts of documentation, and AI can help summarize information or flag things that might need a second look. Engineers use it when they’re digging through technical literature or thinking through design questions. It can be a helpful starting point. But nothing moves forward without a person reviewing it. We treat AI as a tool that speeds up early thinking, not as something that replaces the judgment of the team.

Joanna’s AI Thesis for Medtech

7. A lot of software people assume progress is mostly a matter of better models and more compute. In medical devices, what are the real constraints that slow adoption, even when the AI looks strong on paper?

Healthcare has a much lower tolerance for uncertainty than most other industries. A model might perform beautifully in a controlled dataset, but that doesn’t necessarily tell you how it will behave across different hospitals, patient populations, or unusual edge cases. There’s also the regulatory reality. Medical devices have to go through structured validation to show that they are safe and effective. When a technology involves AI, you have to think carefully about how that system behaves over time and how you test it. That process takes time, but it’s there for a reason.

8. In consumer products, AI failures often mean users leave. In medicine, failure can harm patients. That changes the math. How does that asymmetry influence the way you judge AI claims in medtech?

The stakes are simply different. In most consumer software, if something doesn’t work well people stop using it and move on. In medicine, a mistake can have real consequences for a patient. Because of that, I pay much more attention to the evidence behind a technology than to the sophistication of the model itself. Impressive demos are easy to create. But what really matters is how the system performs in real clinical environments and whether there is strong data to support its use.

Joanna’s Vision for What Comes Next

9. There is a lot of excitement around AI diagnostics and clinical decision support. From where you sit, what is overstated, and what is underestimated?

I think the idea that AI will quickly replace clinical judgment is probably overstated. Medicine involves context, experience, and subtle signals that are difficult to capture in a model. Physicians are synthesizing information from many sources at once, and that kind of reasoning is not easy to automate. What I do think is underappreciated is how helpful AI can be behind the scenes. Tools that help clinicians interpret imaging more consistently or organize complex patient data could improve care without needing to make the final decision itself.

10. And to close on something more personal: given that failures in this field are not reversible the way they are in software, where do you personally draw the line on acceptable uncertainty when AI is involved?

There will always be uncertainty in medicine, but technologies that touch patients need to earn trust before they are widely adopted. For AI, that means strong validation and a clear understanding of how the system behaves in real clinical settings. Personally, I’m cautious about letting AI influence clinical decisions unless there is solid evidence that it improves outcomes. Until then, I see the greatest value in using these tools to support the people building and delivering care rather than putting them directly in charge of it.

Origianl Creator: Ekaterina Pisareva
Original Link: https://justainews.com/industries/healthcare-and-medical/interview-joanna-nathan-ceo-and-co-founder-prana-surgical/
Originally Posted: Mon, 16 Mar 2026 13:00:15 +0000