Bonus episode: The role of AI in healthcare, more specifically pandemics
A pandemic can start with a picnic. But can we prevent pandemics without cancelling feel-good gatherings? Enter artificial intelligence (AI). Dodi meets up with an AI expert panel and finds examples of the real healthcare potential of this trendy topic.
CONOR : Hi guys, it's Conor here. I hope you've been enjoying the podcast as much as Dodi and I have enjoyed making it so far. Look, we've got a whole bunch of new and interesting stories in the pipeline that will be coming to you soon, but before that we've got a bit of a bonus episode. Earlier this year Dodi was at SXSW in Austin, Texas, and she got the chance to sit down with some leaders from Qlik, GE Healthcare, Magic Leap and Vulcan and talk a little bit about the role of AI in reducing the impact of pandemics. I hope you enjoy it and we'll be back soon with more Discovery Matters.
DODI : What are your observations so far about the conversations around AI?
DIRK VOELKEL : There's a lot of talk about AI here at the conference, of course, I mean there's basically in every panel and the discussions are really kind of techy focused on one end, and what’s possible with it, what can we do. We just went to session where people were talking about what happens if you upload your brain up into an AI and you’re living forever, but then there's also a lot of discussion about the social implication. So, people will lose their jobs, what's going to happen? Are people displaced? What are our social responsibilities as companies or as politicians and society? It's very broad and I think our topic is going to be healthcare. Which is really fascinating as it’s already used there to help people. So, it’s everywhere and it’s really fascinating. And my name DIRK VOELKEL and I'm working for GE [Healthcare] Life Sciences.
JULIE K (WHIPPLE) : Just like Dirk said there's a lot of technical talk around artificial intelligence and I think what’s really exciting about our panel and all the representatives here, it’s about like, using AI already and for a lot of us that's kind of just the power of data and analytics. Looking at historical data and how that can help with kind of models, and engines to actually predict and/or identify priorities for the future. My name is JULIE K (WHIPPLE) and I work for QLIK.
DODI : How is society thinking about artificial intelligence?
SUMI PARANJAPE : Well I think that one of the things that’s interesting from the panels that I've been on is, as you said, there's great deal of breadth, but many of the data sources that people talk about using are the same throughout different talks whether it's a talk about oceans or talk about health. So I think there are a lot of synergies there, people talk a lot about the sustainable development goals, 1 through 17. And again, we have siloed those so that we can address those. But one of the powers of AI is that it lets us leverage really diverse sets and disparate sets of data to answer multidisciplinary problems. I’m SUMI PARANJAPE, I work at Vulcan.
DODI : And is there a question that we're not asking about AI that we should be asking?
JENNIFER ESPOSITO : Well, I think in healthcare we've seen a lot of people focus on these really extreme clinical use cases and they jump to these conclusions about replacing physicians. And so, I think we should be asking how we can use AI to assist, to augment what we already have and, you know, I also think about some of these data sources that we’re currently sitting on that's going underutilized that can potentially impact healthcare. I think in the conversation that we're going to have around things like pandemics and infectious disease, there's lots of data that already exists that we could be using better and more real time to really predict things and intervene earlier – versus thinking about sort of the more scary picture of a place where we're trying to do radiology without any humans involved or surgery or whatever it might be. I'm JENNIFER ESPOSITO and I work for Magic Leap.
DODI: So let's get to pandemics and there is the adage that says, you know, a pandemic starts at the church picnic when everybody eats the potato salad. And here we are gathered around each other and it's the coffee, there are cookies on the table. Could it happen here and now - and if it did what would be the role of AI in helping reduce the effects of that?
DV: I would argue it is happening, it does happen…
DODI: And it’s in the cookies right now.
DV: …and it’s in the cookies right now. What I mean is that diseases, they don't have a major impact, right. You get cold, a sneeze or something, but it's happening all the time. We have those viruses living with us all the time. Just we're not aware of it, they live with us. Maybe sometimes we don't have even symptoms, but it's about finding where it’s really relevant. We don't need to know about everything but the real relevant stuff we need to find out early enough so we can actually tackle it as a society and be able to counteract it. Particularly, let me think about it, the flu: when it really kind of gets really bad or if you go to Ebola and some other cases like that.
JK : Back in 2014 data was very relevant when it came to identifying the real urgency around the Ebola crisis in West Africa. Dr. Hans Rosling of Gapminder and of Sweden, of course, looked at the data and recognized very quickly that the cases and death rates actually were doubling at that point in time. And quickly the private sector, the public sector, everyone was springing into action. The most interesting thing I think of when I think about that story is not just necessarily the data being used to create that sense of urgency, so that there was a response - but also, data being used about six months to eight months kind of after that September 2014 timeframe, when they actually use data and Dr. Rosling drove this to identify that what they were doing was in fact working and I think that's just as important. To leverage data not only to identify the concerns and the priorities of where we apply our resources, but also as we are responding: What is working? What is not? And that can be shown through artificial intelligence, augmented intelligence and the use of data. So, it's very powerful.
SP: I think what you're saying is really important and one of the things that we in the public health and global health communities have done for decades is to use data from epidemiologic purposes. And this helps us to understand when an outbreak might be starting and how to respond to the outbreak. One of the powers of AI that can really move us beyond the way that we handled the 2014 Ebola outbreak, is that we can also start to look at predictive analytics and start to anticipate when an outbreak might be starting. So, that those astute clinicians that Jennifer referred to actually have the data that they need at their fingertips to understand: if this is a flu outbreak or is this maybe malaria right now? And then you know beyond that as we integrate data across that spectrum of an outbreak, we can start to reuse many of those data sources.
JE: How you potentially solve some of these problems with respect to pandemics and infectious disease is thinking about the underlying infrastructure and the health system overall and the health of the health system. And I think AI allows us to do all of these things concurrently versus just responding to the epidemic as it happens, thinking about health system strengthening, health as infrastructure. You have all of this other data that you can put to work to really understand where you may be most at risk or where you may need to sort of tweak the underlying system to…
DODI: I'm starting to feel sorry for AI. Those are enormous expectations.
JE: Well, the other thing I think is really useful from an AI perspective is every time we've seen these outbreaks occur, there’s been sort of a massive effort to get supplies, vaccines, needles and gloves to where it needs to go. And I think that's another place where this can really help just from the logistics, and the supply chain perspective. There's a huge opportunity there to make sure that you're ready, number one and that you're getting what needs to go where at the right time.
DODI: So, a bot could be ordering the gloves that need to go on that airplane with food and medical supplies.
JK: And that's already happening today. We work with an NGO out of Santa Barbara California called Direct Relief and they’re the humanitarian supply chain accepting donated products from several pharmaceutical large companies around the world and they combine them on pallets and ship them out to humanitarian. They were very instrumental in stopping the Ebola crisis a few years back. And they're using, kind of, data and analytics around all of their past responses to be able to predict how many Band-Aids, how many gloves, how many suits are going to need to be on the ground and they're proactively receiving those goods and actually getting them shipped to the partners on the ground already.
SP: I think another thing to add to that another dimension AI is really gets to the individual. We've always taken population health approach to particularly the infectious diseases. And now we have an opportunity to look at the factors between us, whether it's our genetics, whether it's immunologic factors that really predispose to getting an infection. I get infections like that. Like don't, just don't even breathe on me. Whereas my husband is fine, you know and being able to target healthcare at different sub sectors of the population is going to be a huge advantage not just for epidemic prone diseases, but I think for healthcare in general.
DV: Just think, I mean just this morning that was in a report on TV about the measles infection how it’s spreading and actually exploding right now, right? I mean, they're kind of like, we already at almost at the number that we were for the full last year for this year. So it’s something that’s not just far away and somewhere in Africa, but it’s actually happening here as well. That's a really important part as well.
DODI: I love that, because I think we say the word pandemic and people automatically go to faraway places. We’re right now in the States in Austin, Texas. So I'll say that Africa is far away and people do not take pandemics personally. How should people start taking pandemics personally?
JE: Certainly things like air travel make the spread of pandemics outside of individual countries very easy. But I think the other thing to think about is the broader economic impacts that happen, not just in that country, but in the world and to the global economy. So, tourism is impacted, ports are closed, people are unable to do export and import. So, I think we definitely have to think about it in a broader context because these things have the potential to really number one spread very quickly, but really have broader implications, not just from the perspective of health impacts.
DV: There's actually another factor that makes it very personal: like the decision to get vaccinated or not with your children. I mean, you're impacting it with personal decisions, right? Not just impacting, like how you're affected but how the society is affected, and I think that's another very precise aspect of the whole thing.
SP: And actually going to that issue around risk communication and general community distrust that we see, unfortunately increasingly around science in general. You know, I think that’s one of the big challenges right now. Not just with the measles outbreak in Washington state but in the DRC with the current Ebola outbreaks. And you know, one of the things that we’re actually launching yesterday, we launched an Ebola response accelerator challenge. And one of the objectives that we have for that is to try to understand how we can better use data and analytics for things like risk communication. How can we start to understand this source of that community distrust so that we can actually respond to it effectively.
DODI: I would just like everybody to lean in here and think of one word that explains why you feel so optimistic about AI and reducing the effects of pandemics. Just everybody, just say the word all at once.
ALL: Potential, efficiency, prioritization, interrelatedness.
DODI: Awesome. Now what you fear?
ALL: Privacy, bad decisions, infrastructure, intransparency.
DV: AI, per se, is not transparent. So that's where the…
SP: …the bad decisions… Yeah, exactly!
DV: So people cannot follow the decision making and that scares people and I think we need to work on that.
CONOR: Our executive producer is Andrea Kilin. Discovery Matters is produced in collaboration with Sound Telling. Production and music by Thomas Henley.