Medical researchers are using artificial intelligence to identify at-risk patients, predict issues before symptoms arise, and potentially save lives. Collaboration between public and private sector, for example, is leading to groundbreaking new developments to images and detect patterns at massive scale.

The following transcript has been lightly edited and condensed for ease of reading.

 

Steven Gray: Good morning, everybody. So, obviously AI is opening up tremendous opportunities across sectors, including healthcare. There’s lots of questions that have been unresolved about how we can really use AI to improve our health. Just yesterday, the FDA announced that Zebra, an Israeli company, has been approved to basically use AI to screen x-rays which could potentially help quicken the speed at which we can diagnose. There’s lots of other research being done around the world on AI and healthcare.

And so I want to invite two awesome researchers who are on the frontlines of that stuff. So Bambi and Terry, come on up.

Bambi DeLaRosa is the healthcare artificial intelligence principal investigator at Micron. We’re going to get into the details of what your sort of scope of work is in a second. But fundamentally, Micron is a company focused on memory. Historically, it’s produced flash drives and, in fact, it’s the world’s second largest producer of, I think, chips after Intel. Is that accurate?

Terry Lo is the head of the Knight Center for Cancer Research—

Terry Lo: No, I’m not the head. I’m a research strategist at the Cancer Institute.

Gray: Okay, thanks for the clarification.

[LAUGHTER]

A slight clarification, at the Oregon Health and Science Institute in Portland.

Lo: The Oregon Health & Science University, yes.

Gray: Thank you.

Lo: The Knight Cancer Institute at Oregon Health & Science University.

Gray: Awesome.

[LAUGHTER]

Thank you, sir. We’re all human here. But, basically, a key part of your work is to develop partnerships between private companies like Micron, correct?

Lo: Right.

Gray: And you’re also a biochemist. Welcome to Techonomy.

Lo: Thank you.

Gray: So Bambi, I want to sort of set the table a little bit. Give us a sense of your background and how and why and when you began thinking about AI and its potential to improve our lives through healthcare.

Bambi DeLaRosa: Yes, so, my whole life story in one minute, “Go.”

Gray: Well, 30 seconds, 30 seconds.

DeLaRosa: So yes, my background is neuroscience. I started with I got a PhD in cognition and neuroscience and I was really motivated in studying the brain because I was interested in solving complex problems that have a really high impact. And in studying the brain, I really think saying it’s complex is maybe even an understatement because when you think about the brain, there’s about 85 billion neurons and each neuron has anywhere from 1–10,000 connections. So, one cubic millimeter of brain volume has more connections than the stars in the Milky Way Galaxy. So, highly complex, which lends to machine learning or AI.

So machine learning is very data hungry, data driven, so when you’re in these complex spaces where you have lots of data and you want to distill this down to particular information, machine learning is a great space. So that’s where machine learning and my research on the brain kind of merge.

And then on the flipside of that, there’s been a number of advancements in AI or machine learning because of what we know and we’ve learned about the brain functioning, there’s spillover effects into AI. So there really is a synergy between the two.

And then, so that’s where my career started, then I had the opportunity to work in the federal government and I was able to work with some really great individuals and we pushed broad AI initiatives. But what this facilitated was an opportunity to really see the landscape from a 30,000-foot view and what I saw was that hardware was really the pulse of innovation. So, that’s where I wanted to be. And it was a place where I really saw potential impact, especially going into the future with AI and machine learning. So, Micron.

Gray: That’s awesome, a little over a minute, but that’s okay. Give us a little bit more context about Micron and sort of how and why the lightbulb started going off in the heads of executives there that it was time to shift away from just being primarily a chip maker to getting into AI solutions for healthcare and, specifically, for cancer.

DeLaRosa: Well, so broadly speaking, I think it’s helpful to give a context of Micron. So Micron has been a leader in the semiconductor industry for over 40 years. It’s a global company that was started out of Boise, Idaho, but has a broad footprint in 17 countries, over 35,000 employees. So the focus is on memory chips. And in the space of AI, you know, AI developments have really come—you’ve probably heard this story a million times, but it really is an important piece, that AI advancements have come from three pieces, one is hardware advancements, the machine learning or AI algorithms themselves, so a better understanding how machines learn, and then data.

So all of these kind of coming up in a parallel fashion has created the space that we’re in now and where Micron comes into play and really is an integral piece of the equation is that memory is where the data lives. So memory is a true facilitator of a number of the AI applications. So memory is in everything, from our phones to the data center to your laptops and the particular specifications of what you need from memory is different and so really understanding the AI space, especially as it’s evolving, is a priority for Micron to ensure that we’re delivering the memory needs that are going to be necessitated in this space.

Gray: And you’re essentially in a research and development team, is it focused across firm or in a specific line of business or what?

DeLaRosa: So I’m in the advanced computing solutions group so we are primarily—we do a number of research activities. And so my space is the intersection of healthcare and AI but we have a number of activities and a broad spectrum of automotive and it covers a broad spectrum of things.

Gray: Terry, at the Institute in Portland, what’s like your core mandate or the core sort of solution or problem that you’re trying to find a solution for?

Lo: So Oregon Health and Science University is Oregon’s academic medical center. The Knight Cancer Institute is the cancer center within the university. It’s called the Knight Cancer Center because we were endowed by Phil and Penny Knight, with Phil Knight being one of the founders of Nike. Our mission is to end cancer as we know it. Our center director, Brian Druker was instrumental in helping to bring one of the first targeted personalized therapies to market, that is Gleevec. And some of the leukemia patients that were initially on Gleevec have been living normal lives for over 10 years at this point.

So we want to find better ways to tackle personalized medicine using the advanced technologies such as artificial intelligence.

Gray: And you’re looking at cancers broadly, not a specific set of cancers, correct?

Lo: Right. We have particular clinical trial programs in particular cancers but we eventually want to solve for all cancers.

Gray: Give us a sense of how the partnership came to be and how it works.

Lo: So we have a long standing need to bring different types of technologies into our research. My role at the Cancer Institute is to work with our researchers to figure out what new technologies, including information technologies and AI, are needed to advance the research. We have—I make input into whether we should buy, build, or partner for those solutions and in cases where we want to partner, we reach out to companies and try to form relationships that are mutually beneficial. We want to help to solve our research problems and also help to advance the technologies and the understanding of our technology partners.

Gray: Can you give some examples, like, a concrete tangible example of how it’s working on a day-to-day basis? Like, one specific problem that you’re actively engaged in solving.

Lo: So one of our big initiatives at the Cancer Institute is something called the Smart Clinical Trials Program. The Clinical Trials Program is different than traditional clinical trials. In traditional clinical trials, the point, or people participate because they hope that their participation will benefit future generations, they don’t necessarily expect to benefit themselves. In our clinical trial program, we are collecting an unprecedentedly deep set of data of all different data types on each individual patient, we are putting in a lot of effort and collecting huge datasets on those patients and we’re bringing a lot of experts from different fields together to understand all the data and inform the best course of treatment for that patient.

Then we are following that patient over time as their cancer progresses on the treatment and course correcting with a new set of analytics at each point where the cancer is changing.

DeLaRosa: And then I can speak to kind of on the technology side, as far as engagement on the day-to-day. Micron is really invested in ensuring that it is a true collaboration. So we sit down with them and try to understand the analytics pipeline and find where the pain points are because sometimes you’re speaking two different languages so it takes a lot to really sit down. So we’re not just creating solutions and then passing it off and saying, “Here you go,” it is helping build memory-centric solutions catered around their particular problems. So sitting down with them and understanding where the pain points are from a technological perspective to help solve that so they can continue on with the precision medicine initiative and we, in turn, develop the next generation solutions.

Lo: And since are a medical university, we have a really hard time getting computer scientists to join us. It’s really hard to hire AI specialists, unsurprisingly, because—

Gray: Because there’s a shortage, or what?

Lo: Because we cannot pay as much as typical technology companies or provide the financial benefits of joining a new startup. You really have to be invested in the mission to want to work where we are. So by partnering with companies with Micron, we get that opportunity to tap into that AI expertise that we might not normally be able to access.

Gray: When we spoke on the phone last week in a prep conversation, both of you mentioned that there are some blockages in sort of the cancer research or in the collection of data that AI is helping you break through. Can you explain what’s happening there and how AI is sort of helping alleviate that problem?

Lo: Sure. In our clinical trials program, in order to analyze the huge volume of different data types that we have for each individual patient and get the data back to our panel of experts in a reasonable turnaround time which we think is 30 days, we couldn’t normally do that without technology facilitation or AI. So some of those data types just take—if you had a human sit down and look at the different images and figure out what’s actually in there, it can take months of time. So we’ve done that for a few use cases and for one of our projects with Micron, we hope to use the AI to speed that time up so that we can do it in a couple weeks and make that 30-day deadline.

DeLaRosa: And then kind of on our end, it’s a good example of research centered around real world applications. So a number of times when you’re developing technologies, you have these use cases that may not be representative of how people are really using the devices in real world. So in our context, an example is if you’re assessing kind of quality of machine learning algorithm, the available datasets sometimes—the resolution of the images may be 128×128 in pixels. Whereas when we go and look at one image from the electron microscope, they are 4K resolution. So, the memory demands in those two contexts are qualitatively different so it really helps us develop tools that are applicable in a real world scenario, so it has real impact.

Gray: But let’s talk about the impact. What’s the end goal for Micron? I mean, are you ultimately aiming for a specific product or a set of products, what’s the vision here, where does this lead to?

DeLaRosa: Well, for Micron, this really—it centers, it follows under Micron’s vision, in general, that is really transforming how the world uses information to enrich life. So, broadly speaking, activities like this unify all of what Micron represents. And then for the research specifically, it helps us understand that we’re providing the memory needs for today, but also giving us kind of a first-hand look at what will be the anticipated memory needs. So we’re making sure that we’re on the frontline of developing the next generation technology, given that we’re looking at really rich, complex datasets, because that’s what it’s going to be, especially in the machine learning world. So it is focused on the research and the understanding of what this space broadly looks like.

Gray: Yes, I want to open the floor to questions from folks who are probably much more informed about this than I am. Any questions for Bambi and Terry? Please, introduce yourself, there’s a microphone coming.

Vansalles: I’m Andy Vansalles, I litigate cases for aggrieved individuals. I’m curious as to what kind of feedback you may get from the general medical community with regard to the applicability in the broader medical world of the work that you’re doing with real individuals now.

Lo: Sorry, can you please repeat that?

Vansalles: What do doctors say about the results you’re getting?

Lo: So for right now, for our particular use cases, we are enrolling only really advanced cancer patients in our clinical trial program. So these patients really have no other options and so far we’ve had really promising results. So we haven’t had any negative feedback. I mean, for our particular program.

Gray: Another question from the audience? Cool. Anything you want to add before we go, we’ve got 45 seconds, roughly.

DeLaRosa: Thank you.

Lo: I’d like to thank our donors, our patients, and our scientists for the opportunity to represent our cancer center for our audience today. Thank you.

Gray: Great, thanks so much, Bambi and Terry.