BeekeeperAI

Microsoft, Intel, and other data giants are buzzing about BeeKeeperAI. A secure collaboration platform, BeeKeeperAI (BKAI) is accelerating scientific discovery and generalizable innovation by helping reduce time, costs, and security concerns for data stewards and algorithm owners.
‍
Michael Blum, MD, recently stepped out of his academic healthcare leadership position to become CEO of BeeKeeperAI as it enters a new phase of growth and spins out of its incubator at the University of California San Francisco’s Center for Digital Health Innovation (CDHI). Over the past decade, Dr. Blum has served as UCSF’s Chief Digital Transformation Officer and the Executive Director for the CDHI.

A clinical cardiologist who was trained as an engineer and founded a health tech software startup during his residency at Yale, Dr. Blum has been embedded in digital transformation in healthcare since the early 1990s. His engineering mindset, subject matter expertise from clinical practice, and understanding of the compute power required to make broad data access a reality give him unique credibility in the emerging field of healthcare AI.

After 30 years of leading health tech innovations, what’s so special about BeeKeeperAI (BKAI) that Dr. Blum would leave his current posts to get it to market? How did he go from chemical engineering to curing cancer via computers? How will AI and confidential computing materially impact patients around the world? He answers these questions and more in an exclusive interview with The Navigator.

Navigator (N): Let’s start by talking about your professional background. You majored in engineering before going into med school. As a young twenty something, what was the connection there for you?

Dr. Michael Blum (MB): I was a chemical engineer in college. I always loved science—it made complete sense to me. When I was young, I loved taking things apart and trying to make them better. Sometimes it worked, sometimes it didn’t, but I was always learning. So, engineering really appealed to that part of me. As I was getting into my junior year, I was starting to get recruited by petrochemical companies, and I had to start making some career decisions.

And two things really struck me. One is that I couldn’t see myself working on an oil rig or inventing a new kind of plastic. Two, the concept of building things that impacted health and lives really was appealing to me. My dad was a doctor, and he took care of people with blood cancers, including kids. And I saw the sadness when he couldn’t help them, but how he impacted their families by trying, by making the effort to always to do his best, and to be there for them in the very hard times.

So I grew up hearing his stories and watching how it lit him up when he helped people. I think without realizing it, that became a part of me early on. So as I was making this career decision, I was like, okay, well, I can go down that path where there’s this really cool new stuff like biotechnology and genetic engineering that’s just getting invented, and I can be a part of that…or I can be a chemical engineer. Without realizing, I had made a decision already of where I was going, and that led to medical school.

N: But it seems like you couldn’t escape your engineering mindset.

MB: True! In fact, I was already creating a non-traditional path during the summers while I was in medical school. I was working in biotechnology companies and experimenting with these very new technologies that were coming along and really enjoying that. This was also very early in the information technology revolution. We were moving off of DECs, minicomputers, and mainframes and onto personal computers. Early versions of Microsoft Windows and powerful databases were just starting to come to market. And so we were starting to build programs that were helpful on individual scales and for small groups, whereas it took huge resources to do that before in very large computing environments.

As I finished medical school and went into my residency and fellowship, I was bringing all three of those things together. The medical education, the clinical training, and then creating new things, looking at biotechnology and how to develop new molecules. New diagnostics was a big part of what I was doing. And so was the information technology, the computing piece of it.

N: When did you cross the divide and start bringing the medical, engineering, and technology parts of your experience together?

MB: It was actually before I left residency. This was very early in the Electronic Health Record era. They were starting to appear in hospitals, but they were really only displays for lab results reporting and better billing systems at that point. When I got to Yale for my residency, they said, “Hey, we see that you’re from NYU, we’re putting in the same computer system they had there—can you help us do that and tell us what you learned?”

It was terrible—there was no graphical interface, useful reporting, or order entry. As interns and residents, we would look up our patients in the EHR and then painfully copy all of the information down onto paper. It was incredibly inefficient, boring, and error-prone. I began trying to hack the new PDAs (Palm Pilots, Sharp Wizards, HP Handhelds) that were coming out at the time and getting the data to download into them, but they really didn’t have useful software development kits (SDKs) at the time.

That was the very beginning of what became the career path of a Chief Medical Information Officer, which didn’t exist at the time. There were very few doctors who worked professionally with the information technology team and could speak technology, appreciate the technical and administrative realities, and explain to them what the clinicians work was really like and what they needed. So that began that that part of my career. It fit in this engineering component and my love of research and development. The work was really asking: How do you take those technologies, with all of their limitations and possibilities, and impact patient care?

As I went through my clinical training, I fell in love with cardiology and stayed at Yale for my cardiology fellowship. At the same time, computer programming capabilities and programming languages were getting much more mature. People would come to me and say, “We’ve got this computer problem in the electrophysiology lab or the Emergency Department, or the training program and we need software to help us, but there is none. Can you help us out here?”

So I built an application that started tracking resident training and performance and another one to track ED activity, and another for the EP group. Next thing I knew, I had a startup that was building software and selling it to training programs. And as I was finishing my cardiology fellowship, I had to make a very difficult decision: Am I going to go be a software entrepreneur and build a big software company? Or am I going to take care of patients like I’ve been training for the last 11 years to do?

N: So what did you decide?

MB: I decided to pursue clinical medicine and take care of patients… but also to create a pathway that kept me in the technology space. I was very fortunate to have leaders and mentors who understood that while the plan was very atypical for an academic cardiologist at the time, they appreciated that data, information, and technology were going to be an increasingly important part of healthcare.

I closed down my software business, but once you’ve been bitten by the entrepreneur bug, and once you’ve started doing things that are outside the traditional realm, you end up always blazing new trails and creating new things. I ended up combining the two careers at Yale and becoming not only a faculty cardiologist, but also the first ever Medical Director for Information Systems. Several years later, I was recruited out to UCSF.

At that point, the world was transitioning into the cloud-social-mobile swing of the digital age. I realized that there was a whole technology and consumer world out there that is moving faster than healthcare and expecting more than ever. We were starting to strategize: How are we going to do digital transformation in healthcare? How are we going to develop new technologies that really get to the patients? How can we help clinicians and patients choose which technologies are reliable and beneficial and which are not? How are we going to realize new artificial intelligence capabilities within healthcare?

And that gets us to where we are now: A full circle from starting a small software development company during my training to spinning out a company from UCSF to accelerate global healthcare AI development. In what seems like the blink of an eye, I’m the CEO of a tech company again.

N: I think entrepreneurs everywhere can relate to that feeling of “Somehow, I just keep starting things!” In fact, six years ago, you co-founded UCSF’s Center for Digital Health Innovation. What types of projects does the CDHI take on?

MB: You know, the clinical space moves so quickly and slowly at the same time. There’s so much need for technology to really help that move forward. At UCSF, we created the Center for Digital Health Innovation to incubate all this cool stuff and put it out into the world. The Center takes the most promising parts of new digital technologies, wearable sensors, new apps—and rigorously validates whether they actually work or not.

We learned quickly that what we call the ‘metabolic rate’ of healthcare and of digital technologies is dramatically different. Every six months, these technologies were being refreshed and completely changed. It was really hard to build things and integrate programs around them at the pace that healthcare works.

If you’re going to develop these programs, you have to think differently than healthcare traditionally does. You have to find things that move much faster. When the AI revolution started to get a toehold in healthcare, we were fascinated. It turns out that if you have a dataset and a powerful computer, you can get moving very quickly in developing machine learning algorithms. A bunch of model projects and proofs of concept had gotten a lot of press: machine learning models, algorithms that could diagnose a bunch of conditions…There was clearly a lot of opportunity in AI.

N: How did the Center come up with the BeeKeeperAI platform?

MB: Healthcare is struggling. Patients are struggling; they aren’t benefiting from technology that already exists because the data is inaccessible. Healthcare orgs are struggling operationally and financially because they don’t have the efficiencies and the capabilities of the AI technologies.

Even when you have high-quality algorithms that show very good accuracy on the dataset you developed them on, that doesn’t mean they’re generalizable across everyone’s data. We could make one that worked really well on San Francisco Bay Area patients. But for that algorithm to work nationally, and globally, you needed a lot more data, and access to data that you generally didn’t have.

This is a global problem. We need to figure out another way to be able to access data in order to really power the AI revolution in healthcare. We need to develop a platform that does it. That’s how [BKAI] started.

N: You’ve spent the last 30 years in professions that are built on big, bold moves. How would you define what it means to be bold in the converging field of healthcare AI?

MB: Part of being an entrepreneur is seeing where things could be, or should be, or need to be. In healthcare, that’s really easy because there’s so much opportunity for improvement. How are we going to use data and technology to really advance care and advance patient experience?

As a young engineer, they teach you how to think about problems, how to break down problems, how to create new solutions to problems. It’s a very structured way of thinking. When you put a clinical experience on top, which is sometimes the opposite—a very human undertaking—and then throw some technology on top of the pile, you’ve got all the components of a breakthrough approach.

We’re in this incredible time where we have such powerful tools to improve healthcare and impact people’s lives in ways that we could never have imagined a decade ago. We’re getting to a place where we can really make predictions and say “Hey, this is what’s going on with you, personally, and here’s what will happen with X or Y treatment.” Based on these incredible new algorithms, the data becomes information—becomes wisdom, even—in our clinical conversations. And BeeKeeperAI is getting us to that place faster.