Dima Romashin, Founder and CEO of Efferon – Interview Series

Dima Romashin, Founder and CEO of Efferon, is a healthcare entrepreneur who has spent the past decade building the company from an early-stage startup into an emerging European medical technology firm focused on treating sepsis and septic shock. As co-founder since 2016, he led Efferon’s business development and growth strategy before becoming CEO in 2026. Under his leadership, the company has developed multimodal hemoadsorption devices designed to remove endotoxins and inflammatory mediators from the bloodstream, helping critically ill patients suffering from severe inflammatory conditions. Efferon’s technology is now used in more than 40 countries and has been applied in over 25,000 treatments, while the company recently achieved a major milestone with CE MDR certification for Efferon NEO, the first hemoadsorption device in Europe approved specifically for neonatal and pediatric sepsis patients. Romashin continues to drive the company’s international expansion, manufacturing growth, and mission to improve outcomes for one of the world’s deadliest yet often overlooked medical conditions.

Founded in 2016, Efferon is a medical technology company specializing in extracorporeal blood purification therapies for patients experiencing severe inflammatory responses, including sepsis, septic shock, acute kidney injury, and complications following major surgery. The company’s proprietary hemoadsorption platforms are designed to selectively remove harmful molecules from the bloodstream while preserving essential blood components, providing physicians with an additional tool to manage life-threatening conditions in intensive care settings. Headquartered in Europe, Efferon combines research, manufacturing, and clinical expertise to develop next-generation blood purification technologies, with a growing international presence across hospitals and healthcare systems worldwide. Through continued innovation and regulatory achievements, the company aims to expand access to advanced critical care treatments and improve survival rates for some of medicine’s most challenging conditions.

You co-founded Efferon in 2016 and spent a decade leading business development before becoming CEO this year. Looking back on that journey, what were the biggest scientific, regulatory, and commercial hurdles in bringing blood purification technology for sepsis from concept to more than 25,000 treatments worldwide?

On the regulatory front, the big shift was moving from MDD to MDR. Regulators started paying a lot more attention to clinical evaluation — you simply can’t get a device approved on good faith anymore. You need real clinical justification, which means more money, more time, more everything. The whole process became two to three times harder, slower, and more expensive almost overnight. When MDR first kicked in, the backlog to even get reviewed was insane.

On the science side, hemoadsorption is still a pretty young method — only about 15 years old. That’s nothing when it comes to getting global adoption and winning over the medical community. We’re genuinely still at the early stages of this technology, which is why so much of what we do is just educating doctors so they’re ready to actually use it.

Sepsis remains one of the leading causes of death globally, yet it often receives far less public attention than diseases such as cancer or heart disease. Why do you think sepsis continues to be underrecognized, and what needs to change?

The more time you spend working on sepsis, the more you realize it’s really a problem about the immune system — it’s not one disease, it’s the whole immune system breaking down. Sepsis happens when organs start failing in a cascade because the immune system is overwhelmed. Honestly, we’re sitting on an epidemic of immune dysfunction and most people have no idea.

It’s also just a much messier problem than cancer. With cancer, we know what’s happening — there’s a tumor, you go after it. With sepsis, it’s way less clear what’s going on or why it happened to this particular person. By definition, you’ve already got at least two organs failing, so there’s no single thing to target — it’s the whole body spiraling at once. And it’s strange when you think about it: people start worrying about cancer in their 40s or 50s. Nobody thinks about sepsis until it’s already happening to them.

Efferon is using AI across both manufacturing and R&D. Can you walk us through the specific ways AI is helping optimize the design, testing, and production of blood purification cartridges?

We go to hackathons just to poke around and see what AI functionality might actually be useful for us. A lot of our work is pulling together scattered data — unifying it so it’s actually usable in development. That includes going through huge piles of scientific papers, which all need to be parsed and standardized before they’re useful. We also use AI to analyze molecules — to figure out where our method could complement a particular molecule, or even potentially replace one.

Then there’s the more mundane operational stuff — AI helps with reporting, sales, marketing. In manufacturing, when we’re building a new facility, basically every blueprint and plan runs through AI now — it’s become our de facto lead construction engineer. And for quality control, we use it to spot which batches are deviating from the norm and whether there’s a pattern worth investigating.

Developing medical devices traditionally involves extensive laboratory testing and iteration. How has AI-based simulation changed the way your team evaluates new cartridge prototypes before they reach the physical testing stage?

Honestly, we haven’t used AI here yet. We do simulate molecules with AI, but those tools are pretty narrow in scope. What’s missing — and this is actually something the Zuckerberg family just committed funding to — is a full physical model of the human body you could simulate against.

The real bottleneck in developing new treatments isn’t finding candidate molecules. It’s testing. Everything eventually has to go through human trials, and that’s where AI just can’t help — you can’t simulate your way around testing in living people. If we had decent organism-level models, built from data on thousands of real patients, you could actually run a lot of that testing on the models first.

Your manufacturing facility reportedly uses AI-driven systems to improve production efficiency and quality control. What measurable improvements have you seen in areas such as production speed, consistency, defect detection, or cost reduction?

When we bring AI into production, what it really does is cut down on errors in how materials move through the process — parts coming in, going out for prep, moving along the chain. That makes the whole thing cheaper and faster. But the real value comes from baking that into the design of the production line itself — the fewer mistakes you make at the design stage, the more efficient everything downstream becomes.

Eventually AI could do 100% output inspection on every device — but we’re not there yet, we want to be fully confident in it first. Right now we do use it to track fluctuations in wash/rinse testing data, which is genuinely useful.

One of the challenges in healthcare AI is trust and validation. How do you ensure that AI-generated recommendations in research and manufacturing are scientifically reliable and meet stringent medical-device regulatory requirements?

We just don’t use AI in areas where we don’t have the expertise ourselves to check its work. For us, AI is about making existing work easier and faster — not about generating new ideas we can’t verify. Everything still gets double-checked by actual people. Honestly, we’d probably land on the same insight or result without AI anyway — it would just take a lot longer.

Efferon’s technology focuses on removing endotoxins and inflammatory mediators from the bloodstream during sepsis treatment. How do you see AI helping clinicians better identify which patients are most likely to benefit from blood purification therapies?

Sepsis treatment is heading where the rest of medicine is heading — toward personalization. Sepsis isn’t one thing; there are really different phenotypes depending on where it originates, like sepsis from the lungs versus the liver. Getting that stratification right is key to treating it properly. Right now there are standard test systems, but I think we’re going to see more and more AI-driven systems where you feed in lab values and imaging, and it points you toward the right treatment path.

The recent CE MDR certification for Efferon NEO marked an important milestone for pediatric and neonatal sepsis care. What unique challenges arise when developing therapies for children, and how did AI contribute to that process, if at all?

We’re one of the few companies that actually sat down and thought seriously about children — there just aren’t many devices out there in the world designed specifically for them rather than just scaled-down adult versions. Kids aren’t small adults. You have to be really careful designing for them — different comorbidities, different blood volume, different clotting behavior, and they often react more strongly to certain therapies.

A medical device typically takes three to seven years to develop. We started on NEO a few years back, and at that point AI just wasn’t good enough yet to meaningfully help with it.

Beyond sepsis and septic shock, you’ve discussed the potential for hemoadsorption technologies in areas such as oncology, autoimmune diseases, and rare diseases. Which of these applications excites you most, and where do you see the strongest clinical opportunity over the next five years?

I’ve always wanted hemoadsorption to break free of being a “critical care only” technology. We think we’ve built something that could actually work as a broad anti-inflammatory platform — the potential applications are huge.

Sure, it’s more invasive than taking a pill. But think about what “invasive” really means — is it putting two needles in someone’s arm to clean their blood, or swallowing a pill and putting a compound into your body that you don’t actually understand the mechanism of? Once we get past that perception problem around invasiveness, I think hemoadsorption has a real future in things like non-critical psoriasis, Crohn’s disease — we’re already testing a protocol there — and more broadly in autoimmune conditions and low-grade chronic inflammation. Our biggest competitor, Jafron out of China, is mostly selling into the post-critical stage rather than the critical stage itself, which tells you something about where this is headed.

Looking ahead, how do you envision the convergence of AI, advanced biomaterials, and precision medicine transforming critical care medicine, and what role do you hope Efferon will play in that future?

I think AI’s biggest impact will come once we have real organism-level models that can simulate different patient states for testing. That’s when drug and device development really changes.

Right now, we’re collecting a lot of data through registries and clinical studies, partly because we know that data will eventually be useful for training AI models down the line. Very little data actually gets captured in intensive care — it’s hard, because everything happens in seconds — but we do it anyway, and it means we’re sitting on a genuinely unique dataset. I think that’s going to be valuable to AI companies building simulation systems for critical care. As for us — we’re just going to keep following the latest technology and using it to get products developed and to market faster.

Thank you for the great interview, readers who wish to learn more should visit Efferon.