Hostname: page-component-6898cfcb55-zkb6p Total loading time: 0 Render date: 2026-11-11T10:45:21.965Z Has data issue: false hasContentIssue false
  • English
  • Français

A conversation with Nima Aghaeepour, B.Sc., PhD, Professor of Anesthesiology, Perioperative and Pain Medicine, Stanford University

Published online by Cambridge University Press:  09 November 2026

Clinical Research ForumOpen the ORCID record for Clinical Research Forum[Opens in a new window]
Clinical Research Forum
Affiliation:
Clinical Research Forum, Washington, DC, USA
Rights & Permissions [Opens in a new window]

Abstract

An abstract is not available for this content. As you have access to this content, full HTML content is provided on this page. A PDF of this content is also available in through the ‘Save PDF’ action button.

Information

Type
Q&A with Research Innovators
Information
Journal of Clinical and Translational Science , Volume 10 , Issue 1 , 2026, e198
Creative Commons
Creative Common License - CCCreative Common License - BY
This is an Open Access article, distributed under the terms of the Creative Commons Attribution licence (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted re-use, distribution and reproduction, provided the original article is properly cited.
Copyright
© The Author(s), 2026. Published by Cambridge University Press on behalf of Association for Clinical and Translational Science

This article is part of a series of interviews with recipients of Clinical Research Forum’s Top 10 Clinical Research Achievement Awards. This interview is with Nima Aghaeepour, B.Sc., PhD, Professor of Anesthesiology, Perioperative and Pain Medicine, Stanford University. Dr Aghaeepour is a machine learning and artificial intelligence scientist who studies how to leverage computational methods to address clinical challenges in perioperative care and maternal and child health. He received a Top 10 Clinical Research Achievement Award for“AI-guided precision parenteral nutrition for neonatal intensive care units.” The interview has been edited for length and clarity.

What early influences shaped your decision to pursue a career in artificial intelligence and medicine?

I come from a family of engineers and have always been fascinated by computers. From a very young age, I started writing code and then studied computer science for my undergraduate degree. Back then, I worked on building robots that play soccer, and the goal was that by 2050, they would win against the human World Cup champions. That was fun, but I decided at some point that I should do something more impactful with my life, so I switched to medicine and biology.

How did you become focused on maternal and child health?

When you first go into medicine and biology, you generally study the problems of older people. I did that for a while, but eventually a senior colleague, who later became a mentor, asked me to help with a study about pregnancy. I was brought on to be the data analysis person, and initially I thought, “Sure, this is data, like any other data.” However, through that project I began to understand that every pregnancy or every newborn saved creates 80–90 years of life for society – and that was in stark contrast to the end-of-life problems I was working on at the time. Because of that study I became fascinated with the application of AI in maternal and child health. I also became very interested in how AI can be used in high acuity care, such as neonatal intensive care units (NICUs), operating rooms, and ICUs.

What interests you about the application of AI in high acuity care?

These are places where dramatic shifts in patient trajectories happen. In a NICU, an operating room, or the ICU, the situations are complex, the timelines are short, things happen fast, and the consequences are very real. This makes these areas interesting, and it also means these are places where it’s been difficult to deploy AI. If you need to transcribe notes or figure out which billing code is missing, that’s an easier place to use AI. Building an AI that functions in high acuity care is completely different.

Which brings us to the award-winning research…

Yes, this project came about because we were working on predicting outcomes in the NICU. AI models are very good at predicting the future – predicting the weather, predicting which car is going to crash, and other prediction paths like these are all very natural for AI models. So, we asked if we could use AI to predict which NICU patients are going to get into trouble. It turns out, we can. Even better, we can predict which patients are going to get into trouble in the NICU even before they are born. I thought, “This is great – it’s going to change everything!” But when I started talking to neonatologists, telling them that AI could predict outcomes for patients who weren’t even born yet, they were not as excited as I was. That’s when I learned the hard way that an AI model just predicting the future isn’t enough. For the AI to actually change patient outcomes, it must predict the future and it has to tell the healthcare team what to do about it. That’s where this project came from.

Why did you direct your attention to parenteral nutrition?

We started by mapping out all the various interventions in the NICU. When it came to nutrition solutions, we asked questions about how people figure out what goes into the formulas, and we found out that the answer is really complicated. That’s because each formulation for total parenteral nutrition (TPN) is complex and individualized, requiring coordination across multiple clinical teams, including neonatologists, dieticians, pharmacists, compounding facilities, and others. As a result, TPM is the single largest source of error in NICUs, and we realized that this is a place where we could very easily have an impact.

What was the solution you developed?

Our team developed and validated TPN2.0 – an AI approach that learns from years of clinical data to recommend safe, personalized nutrition for patients in the NICU. The system identifies a small set of optimized nutrition formulas and assigns treatments to newborns based on routinely collected data from their EHR.

How could TPN2.0 change clinical practice?

The goal is for the AI-driven standardization to enable mass manufacturing of TPN, which will reduce errors, lower costs, and make this life-saving therapy accessible to newborns in low-resource settings around the world. In the future, we want the neonatologist to see the patient and order, for example, Bag Number 3. Bag Number 3 would already be there in the NICU, ready for use. This would remove the back and forth between the clinical teams, the compounding, the overnight shipping, all those complications. Of course, that will help in NICUs like at Stanford, but the impact will be even greater on lower- and middle- income settings where they don’t have the same kind of infrastructure.

Where is this research headed next? What are you working on now?

TPN is just one example – there is so much more that happens in a NICU. How do clinical teams establish regulator settings or decide on various medications? How do they determine who even goes to the NICU and who goes to what level? These are all problems that AI models will be able to address very well. That’s one aspect. The other aspect of this work that I’m very proud of is that people often say it’s hard to innovate in pediatric spaces because the patient populations are smaller, there’s not enough venture capital interest, etc. But we have proven that to be wrong. We have shown that you can innovate in pediatric spaces, and you can scale. In addition, after we built the system for the NICU, we received inquiries about doing something similar for adult patient populations, and now we are building parallel solutions for stem cell transplant patients, for ICU patients, and so on, expanding this to adult high acuity settings.

You’ve mentioned “teams” a few times. What role does teamwork play in clinical research?

This kind of work can’t happen in isolation. If you look at the author list [on the award-winning paper], you will see pharmacists, dietitians, neonatologists, pathologists, anesthesiologists, and a small army of people who do machine learning and AI. If the goal is to publish papers, you can do that in a more isolated, small environment. But if the goal is to build something that actually gets used, you must make it work for the people who are going to be the users, and that takes a larger team. For this paper, we did pilot studies with different stakeholders and listened to them to learn which elements they didn’t like, and then we went back and made changes. Building the right solutions requires deep collaborations, and that’s a skill that any young investigator in AI, and in medicine, must develop.

How can a researcher develop those kinds of communication skills?

It comes with discipline, practice, and having good mentors. I’m the most introverted person I know, but through practice and by having good mentors, I have gotten to a place logically and emotionally where I understand the importance of speaking with my stakeholders and doing that broadly and frequently.

The use of AI has skyrocketed over the past few years. Has that acceleration surprised you?

Things are moving fast. I’m doing things this week that I did not even imagine doing a month ago. One area that is surprising to me is how much AI is helping with science – and not just biology and medicine, but with scientific research into AI itself. Right now, some of the smartest scientists in our lab are AI agents. I see this as an opportunity to reinvent what it means to be a human, what it means to be a scientist, and where we can create the most value.

What interests do you pursue outside of the lab?

I enjoy many forms of extreme sports: airplanes, acrobatic flying, skydiving, wingsuiting, and skiing, to name a few. Basically, anything that flies or goes fast brings me joy, on a primal level. I have a three-year-old daughter, so I’m less involved with aviation than I’ve been in the past, not because of the risk, just because hanging out with a three-year-old is so much fun. My family and my hobbies help me reset and remember that there is a universe outside of work.