Project Ronin is an oncology intelligence platform company founded by Dave Hodgson, David Agus, M.D., and Oracle co-founder Larry Ellison — to help oncology care teams make personalized, data-driven care decisions. Hodgson, the company’s CEO, recently spoke with Healthcare Innovation about how artificial intelligence is beginning to bolster the day-to-day work of oncologists.
Prior to co-founding Project Ronin, Hodgson had over 20 years’ experience in senior executive roles at both major life science companies (Pfizer, Roche) and fast-growing healthcare startups (One Medical Group, Tethys, Oxeon Partners).
Last October, Providence Saint John's Health Center in Santa Monica, Calif., announced it had selected Project Ronin to implement a cancer intelligence platform to enable cancer patients and their clinical teams to better manage treatment symptoms, prevent avoidable emergency department visits, raise the standard of care, and improve patient outcomes. The company also has published research conducted with clinicians at the University of Texas MD Anderson Cancer Center in Houston.
HCI: First, Kenji Mizoguchi’s The 47 Ronin is one of my favorite samurai movies. What did the name mean to you when starting the company?
Hodgson: It isthe idea of an entity of great power — the samurai — who wasn't beholden to a single feudal master. The feudal master in this case being the conventions of U.S. healthcare. We said let's take a very powerful concept and not necessarily be beholden to how it has been done before.
HCI: Things have been changing fast in the AI world with the rise of large language models. Have the capabilities changed since you started working on this three or four years ago?
Hodgson: Very, very much. A year ago, I would never walk into any provider system and say I have an AI solution. I would never do that. There's no doctor anywhere who says ‘Please have AI tell me what I should do.’ Rather, we said we're using advanced data analytics to serve up some of the choices you might have when considering treatment options or management options for your patient. And, of course, you’re in the driver's seat. I'll tell you where the lanes or the curbs are. I'm not trying to take the wheel from you. Rather, I'm giving you a lot of choices and saving you time from having to fish around to find out, statistically, is this regimen better than that one? Now if I talk to certain groups inside of a broader health system, I say I'm using deep learning, and I have one of the largest clinical language models ever built. We are enabling clinicians using ethical and safe AI.
HCI: Is the idea that you train the models on the health system’s data or do you have a large central oncology database that things are already built on? How does that work?
Hodgson: Both are true. We have built a large oncology data model of our own that we've trained from lots of data. We also built a very large language model in collaboration with Oracle because you need a lot of CPUs to do it. Then when we integrate a new customer, we take their data and we run another training iteration using the localized data that they have. Transparency also is very important. What you have to do very carefully is monitor the performance of your model. You have a training set with which you created your data model, but then you have real-time data feeds monitoring the difference between the training set and the real-world data set. When you present those insights to the clinician or the team, the algorithm is also confessing its known performance. It is saying, ‘Hey, look, I think your patient looks like this set of patients and this is what happened to them. And here's my confidence that I think that's true.’ So you have the transparency and the monitoring.
HCI: The company that I'm most familiar with in this space is Flatiron. Are there some ways that Project Ronin is different from what Flatiron does?
Hodgson: There are three big distinctions. One is that Flatiron is a fully functional medical record system in its own right. We are not. We augment the investment that a provider system has already made in Epic or Cerner. So we plug into Epic so you don't have to buy or adopt us as a full solution. I don't want to build the EHR. I want to extend it and make it better. That's distinction number one. Number two is Flatiron is really at the community oncology practice level, and we are more at the academic medical center and large hospital system level, groups that have massive investments in Epic and Cerner. The third distinction is that their business model is to essentially give the EHR away for free or charge a very small amount because community oncologists do not have large budgets for software. They'll give the software away, but then they're going to take the data and sell it to pharmaceutical companies. We are not selling provider data to anybody.
HCI: Are there specific challenges to breaking into the oncology market?
Hodgson: There are a few specific challenges and there are a couple of specific advantages to it. The big challenges are that in the U.S. provider market right now, a lot of systems are suffering financially, because labor and supply costs went up while their revenue is subject to multi-year negotiations. So a lot of them are getting squeezed. Most providers say I can only adopt something that is going to save me money or make me money. So I say, well, in oncology, I'm going to save you money and make you money. And they’ll respond, but is it just oncology? They are looking to maximize revenue and savings across multiple service lines and we are only in one service line so far. That's one challenge. The other challenge, which is also sort of an advantage, is that oncology is ridiculously complex. And that's why we started this, because cancer actually is hundreds of diseases. No practicing oncologist, as good as they are, can be specialist enough to know everything about every tumor type and every disease option, and that's why we're relevant. We fit pretty well into the complexity of it.
HCI: What should we expect to see from Project Ronin in the near future?
Hodgson: We wanted to build a very powerful product first, so we've actually spent a lot of time in the lab building this stuff. Now we have a comprehensive platform that not only does the AI and is fully integrated into the EMR, we also have a whole patient engagement system. We are also going to give a population health view, so you get a total quality view of your entire cancer program. Now we have all these different components to the platform, and we want to be able to announce in the coming months how many systems are adopting that, and the impact that we're having on patients, which is, at the end of the day, what we want to be proud of the most.