Loading summary
Transcript45 lines
- [00:03]
A
Welcome back to part two of three of our series on AI for Health. You're listening to the Africa Health Ventures podcast where we unpack what's going on in healthcare today and what's going to happen in the next 10 years. I'm your host Rowena Luke. Today we'll be speaking with two guests whose job it is to anticipate the harms to society that could be caused by AI and to stay one step ahead of them. Our first guest is Andy Pattison, Team Lead of the Digital Channels Group at the World Health organization. In the second half of the show, we'll hear from Dr. Sam Ote, senior Program Specialist at the International Development Research center, or idrc, a Canadian funding agency which was one of the first in the world to launch a major funding program specifically targeted at AI for low and middle income countries. But Sam also has a secret alter ego. He is host of the other Digital Health in Africa podcast called the MedTech Africa podcast, which actually aired its 50th episode just yesterday. Because Sam and I have one brain. It's also about AI for health, but in his case he's getting the woman's health angle. You can find it by searching for MedTech Africa, where wherever you get your podcasts. A few quick announcements before we dive in. First, I wanted to give a big shout out to our sponsors, Reach Digital Health, for making this entire podcast series possible. Also, huge, huge thanks to Alice Liu of Baraka Impact Finance, who's just been an immense resource and ally to me in putting together this series. Secondly, if you want to get notified of the next and final episode on our AI for Health miniseries, be sure to subscribe to this podcast. In our third and last episode of the series, we're going to dive under the hood of AI for Health and speak with the implementers and data scientists actually building out novel AI systems for health in Africa. And last but not least, if you want to stay in the know about healthcare ventures in Africa, you can subscribe to our newsletter@africahealthventures.com Newsletter One friendly reminder the content here is for informational purposes only and should not be taken as legal, business, tax or investment advice or be used to evaluate any investment or security. All right, enough with the announcements. Let's get back to the show. We're going to get started with Andy Pattison from the World Health Organization. He's going to tackle all the questions we're asking the WHO about AI, including what are you using it for, what are you scared about, and how are you mobilizing the global health sector to respond to all the risks we're facing today. Here's Andy.
- [02:46]
B
So, I'm Andy Patterson. I work at the World Health Organization. I lead a team called Digital Channels. And our job is to get more health messages to more people through more channels so that they can make better health decisions and hopefully live healthier lives. And that's all about us reaching citizens to give them the best possible health advice.
- [03:03]
A
Fascinating. Great to hear. And, Andy, I know in your role at the who, there's lots of different things that the WHO does. You know, everyone knows the name, but we maybe don't know exactly what it does. Can you talk a bit about how your role relates to other actors in the industry and specifically how AI has affected that interaction over the course of the past year or two?
- [03:25]
B
Yeah, definitely. So the way my team works is on three pillars. The first pillar is to raise good content, whether that's from WHO or Ministries of health or trusted actors in the health systems. The second pillar of our work is to fight misinformation. And the two, pillar one and two go hand in hand because misinformation thrives in a vacuum. And what we'll do with fighting misinformation is not only focus on individual posts or individual myths that need to be debunked. That's the tiny part of the work, actually, because they come out so quickly and so fast and the shelf life is sometimes very short. It's very hard to fight those. A lot of our work is with the tech companies and social media platforms, working with their policy security and safety teams to improve their health policies so that their algorithms and engineers can remove harmful content very quickly and reduce content which is ambiguous or could be seen as misinformation or misinterpreted, and then also to raise the good content. So a lot of the work that we do with those policies and we're coming up with these policy teams, we're coming up to four years now, every week, meeting the big players, whether it's TikTok, YouTube, Facebook, which is WhatsApp, and of course, Instagram, also Chinese company Kaishu, to see how we can help their teams get stronger health policies. So we're coming to four years, week in, week out, we take these calls with them and we flag what we think are problems, and they flag what they've seen on their platforms, and we try to advise them so that they can have better health policy. And then the third pillar of work, which we're very fortunate to have, is gaining user insights to find out what people are doing and also try new tools. And this is where the AI element comes into my area of work. We get to try all the cool, fun tools like large language models and AI and things like that to assess the potential impact for advancing public health.
- [05:22]
A
Fascinating. Let's go through those pillars that you mentioned, maybe starting with the misinformation one you mentioned. Some of the big actors that you work with, can you talk about one of them and just the dynamics of working with them? Anything interesting that you saw or anything that you were able to develop together with one of these big names that we've all heard.
- [05:40]
B
So, I mean, we work across the board and I'll just choose one because it's a simple one and I know the numbers and the numbers are public, so I can talk about it freely. The work we do with YouTube during the pandemic especially was very important to get. I think it was at the end of 2020, nearly a million videos removed from YouTube which contained harmful information. And what we'll do is we'll sit down with these companies and tell them, you know, if we see something that we're a little bit worried about, we'll say, you know, this is the beginning of COVID If you remember going back there, there was all these myths about how to avoid Covid, you know, gargling hot water, gargling salty water, clearing your nasal passages. And some of these are not harmful, right, to the human. And we had to explain to YouTube that actually, although they're not harmful for the human, they make the human believe that they are more resistant to the disease and therefore they take more risks. And it's the risk taking which makes them more harmful. So in these cases, it's a case of showing not that the black and the white is always obvious for these companies, right? There's some things which we can definitely point to and say it's impossible 5G caused Covid. That was one of the big myths. And on the other side, there's stuff which we can never prove because it's so ambiguous, or the science just isn't there yet. But where they need the assistance is in this gray area in the middle, where it's anecdotal, where it's plausible, where there's causality questions with regards to diseases and impact on human life. And we'll talk them through that to sort of show them the background of what's important, what's not, when considering something to be raised or removed. And this is where these conversations come, you know, about eating garlic will protect you from COVID was another one. And of course, if you do that, you're just going to have some bad breath. It might be unpleasant for your partners and people around you, but it's not going to kill you. But that point I made about then you think you're immune and you take risks. That's something that the scientists know, but not necessarily the policymakers at these social media companies. So we explain those nuanced risks and the impact of that knowledge that people have gained to explain to them why it's important to remove that content as well.
- [07:56]
A
You know, the other day I was logging into YouTube and the very first commercial that it showed me was Elon Musk looking exactly like Elon Musk, saying, send me fifteen hundred dollars and I will make you a cryptocurrency billionaire. And it was the same advertising as all the normal advertising that I get on YouTube. And it blew my mind because it looked exactly like Elon Musk. In the era of generative AI that we exist in, you can imagine that's only going to get worse. You're going to see all sorts of public health officials and authorities that are machine generated to the point that you do a video conference call with someone and it speaks in a voice that you recognize and people that you know, does that scare you at all? And like how what are we going to do in a world where that misinformation problem is only going to get so much worse?
- [08:42]
B
We're very concerned about that on two fronts. Not only the ability for deep fakes to mimic somebody so exactly that you actually believe it's them, and we've all seen them and they're very worrying, but also the volume at which and the speed at which those things can be created. So even if they're not as accurate as you want, they could generate a lot of content even without the voice and video. A lot of content can be generated with AI in very short amount of time. And how do we fight that? We've seen some technology companies have put a little label on. I think TikTok the other day said this is generated by AI tiny label that appears at the bottom at the same time. That's only when the creator declares it. So we need technology to do this. The other day, my director, Dr. Professor Alain Labrique, he played some AI voice, some AI voices of some of our colleagues after 30 seconds. And you've seen these, right? Record somebody for 30 seconds and then you get them to say whatever you want and it sounds like them, they stutter, they pause it's very realistic and very scary. So we're hugely concerned. What we plan to do in that is that to convene who has a very big convening power. What we will do is we'll convene the Tech Task Force, which is a group of people I bring together on a monthly basis across the tech industry, and we'll talk to them about some of our problem definitions and we ask them to go off into the real world and find solutions. We're going to bring them, especially on this topic. In June, we're going to organize this big conference. Hundreds of people will be invited from dozens and dozens of tech companies where we will ask them to inoculate the Internet from misinformation. A little bit like a vaccine inoculates somebody from a disease. Let's get really good at stopping going up rather than good at bringing it down. And that's what we want to do. We want the Internet, we want their algorithms and their technology and their thought leadership in the digital space to be able to spot these deep fakes, to be able to spot this content and not allow it to go up and not allow it to go viral. And I know that sometimes it's going to be a negative impact on their business model because volume is views, views is ads, ads is money. At the end of the day, this is a global good and we are saving lives. And so I think that in the past they've been very receptive to these things and I think they're equally concerned. So that's what, you know, our biggest fear with that is the quality, but also the quantity and speed at which it can be done.
- [11:14]
A
I'm just going to say Andy's got a super tricky job. Not only does he need to rein in the worst that big tech can do, sometimes, as he said himself, swimming upstream against natural market forces of billions of dollars or more. But at the same time, he's also working to capture the best of the innovation, the opportunity and the talent that they bring to forge partnerships that can help shape a better society. I respect so much the work that he's doing, and it sounds super hard. In this last bit for Mandy, we chat about user insights, how the WHO is using AI to understand the world and its people better, to understand us.
- [12:03]
B
So we use a combination of traditional ways and AI to gather it, because obviously the AI is still fairly new in public health. It's not been new for who. We've been working on it for five or six years. Looking at how AI can, can help support the advance of public health. So we will use multiple channels. And I think that the use of AI to find out user insights is one of many indicators that you should use. You shouldn't rely 100% on AI until you trust the tool entirely. But we will plug in into ChatGPT and large language models, questions about audiences, about where they sit, what they consume, how they consume it. At the same time, we'll use tools which are out there, like Google Trends and things like that, which are more traditional, based on data of what's happening, to also validate what we're doing on AI.
- [12:54]
A
Are we talking about you, Andy, sitting in front of ChatGPT4 saying, hey, is malaria rising in Mali right now? That kind of thing?
- [13:01]
B
Yeah, member of my team will definitely do that. So if we go back a year when Monkeypox was happening, we were trying to. Monkeypox is a disease which affects everybody. But the most vulnerable community were men seeking sex with men. What we were there, what we did there was to use AI to find out where they were, what their habits were, how often they visited sites where they met each other. And also we went traditionally to these tech companies who have dating apps and hookup apps for that community and said to them, this is the problem definition. We then worked with the industry to get our messages out to that vulnerable population through the platforms that they're on and figuring out from AI to find out where they are, what they're doing, how they're consuming content, and also from the traditional Google search and Trends to figure out where we should be pitching our messages and how. And the industry were actually super helpful in this space because they were able to carry our messages rather than us dragging these people into our social media channels or our website. Our goal is to get our content into their digital journeys and through that mechanism, also the community, somebody is a community that's extremely aware of their sexual health and so took it very seriously. But the next time somebody was on Grindr looking for a date, there was a message from who and the organization that we worked with, the companies that we worked with were able to make that content in a way, in a format, in a shape, with the right tone of voice that this community actually would engage with. So rather than taking our boring, bland Monkeypox advice, which sometimes we have, and WHO is a very scientific and academic organization, only sometimes. Yeah, they would make it cool and interesting and fun and engaging and cheeky as well, which, you know, we can't. They could use all these human side of things, which we weren't able to do.
- [15:04]
A
Awesome.
- [15:04]
B
So the other way we use AI, we have a community called Fides. Fides is a group of 400 healthcare professionals on social media. Every week we send them content and say, this is what we care about at the moment. So in January, we had a cervical cancer awareness month. We sent them loads of content on cervical cancer. This community create content and then send it to their followers. So you have all these humans creating content. The way that some of these people use AI is to get their content translated. So you've seen these tools where you can speak and just change the language and it actually looks like you're speaking a different language. So we have a colleague in South Africa who uses these tools to reach vulnerable communities. And that's another way that I think AI can really help is in the personalization of the message, whether it's just purely language translation, like I've explained, or actually making it much more personable, customizable to the person's age, gender, race, religion, sexuality. And I think that's where it gets very exciting, where the AI can change just the nuance of the language, maybe from formal Portuguese to Brazilian Portuguese. Right. With a southern Portuguese accent to make that a little bit more relatable to that population as an example.
- [16:14]
A
Awesome. What is one innovation or change that might make a big difference to your work in the next 10 years?
- [16:19]
B
I think it's the translation and personalization. I think taking that simple video message that you can produce in English and having it translated on the fly into 60 languages and sent out to vulnerable populations. And if you consider some countries who have literally 60 languages within their country, from a government perspective, that speeds that whole process up, that timeline up. And misinformation can be generated in 30 seconds with a telephone. And science takes weeks and months and sometimes years to research and then to communicate. AI can shorten that time. We stand a better chance of fighting misinformation. So for me, this type of product, which allows quick time to market once it's done, I think is very cool. If I may add another one, by all means. When it comes to social media tools which allow creators to create good content and suggest to creators how we're working, I can't say with who because I'm under NDA with them. But a new tool will come out which will allow medical creators, it will propose to them a bunch of content ideas based on their content and interactions of their audiences and say, listen, in the last 10 months, this is the most popular video this day is coming up. Why do you Want to talk about cervical cancer? And if so, here's three scripts. Do you want to focus on the younger woman, the older woman, or just awareness in general and then depending on what they choose. So it's not just that reaching populations once the communication is there, but it's the upstream designing, building and filming and producing content which is going to be reduced very quickly, reduced in time. And that for me is very exciting. This means that instead of, you know, the average medical creator will spend six hours to create a good two minute TikTok. Six hours for two minutes is a long time. If that can come down to two hours, well, they can do three times as much. And if we can then use the tools to allow us to translate, they can reach, you know, 85 times more people. So I think that's the power that I'm excited about.
- [18:24]
A
Incredible. Can you give a shout out to a nonprofit entrepreneur or innovator who you personally think is doing promising work?
- [18:30]
B
So I'm saying this not because I'm here with the team at Reach, and the reason I came is because I do think that these people are doing an incredible job. I especially like what Reach and Turn IO are doing with regards to Mums Connect, which is my favorite chatbot on WhatsApp, which allows pregnant women and women with small children to get the advice from their government, from doctors and nurses when they need it. Whether it's in the middle of the night, without waiting, they just talk to the chatbot and the chatbot replies. And there are nurses, also humans there, which can supplement the work that's, that's not already in the chatbot, and it reduces the amount of people that need to go and see a nurse in practice, which is one of the big problems is there's not enough doctors and nurses in the world. And in some countries there's a huge gap. So those that really need to see them can see them faster. And those that just have worries, you know, they can just be told at 4 o' clock in the morning by a machine, you're okay. This is normal in your third trimester for this to happen, but if it continues for 48 hours, then contact the nurse. And this allows people to feel a lot more comfortable. And it's a big drive to access health information. I think these people are doing incredible work and I would love to see that replicated worldwide. Because our job is not just to support ministries of health, it's also to get ministries of health to talk to each other so that we can actually learn from each other. And I think that's also key for me is that part of the research agenda with people like Reach and Turn IO is important to them as well. Let's get publications out there to advance
- [20:03]
A
public health despite all the risks and the doom and gloom of AI, Andy leaves us on a note of hope about how there's still a ton of opportunity for AI to reduce the load on an overburdened health system, to increase demand for healthcare information, and to push forward the research agenda so that we can replicate our successes across multiple countries. Lots to think about. Our Next guest is Dr. Sam Otey, Senior Program Specialist at the International Development Research Centre, or idrc, a Canadian funding agency. We wanted them on this podcast because they were one of the first funding agencies in the world to lead a major funding initiative focused on the use of AI in low and middle income countries. So we spend some time in the next part of this podcast talking about how that happened and what other funders can learn from their approach. I gotta say, the other awesome thing about Sam is that he has a secret alter ego. He is also, on nights and weekends, host of the other digital health in Africa podcast which is called the MedTech Africa podcast. That podcast showcases digital health and health tech innovations across Africa. If you want to listen to it, you can just search for MedTech Africa wherever you get your podcasts. But for now, let's start with the IDRC side and their work in AI for Global health.
- [21:26]
C
Here's Sam so my name is Samuel Oti, background in medicine and public health. Started out as a clinician many, many years ago, then decided to make the shift after a short stint in practice in clinical practice. Made the shift into public health and haven't looked back ever since. My current role and my day job is as a Senior Program Specialist in the Global Health Division of Canada's International Development Research center, or IDRC for short. IDRC is a crown corporation that for the past 50 years has been investing in knowledge, innovations and solutions to some of the most pressing development challenges in low and middle income countries. Essentially, we fund research for development that is led by researchers and research institutions in the global South. I sit in our Nairobi office. Our head office is in Ottawa in Canada and our Nairobi office is one of five regional offices. So it is our regional office for east and Southern Africa. Now about our Artificial Intelligence for Global Health initiative, or we call it AI for GH for short. It's a five year roughly $16 million investment in artificial intelligence for health and what we are doing is to fund researchers in the global south to strengthen health systems by leveraging what we are calling contextualized, responsible artificial intelligence solutions to improve global health and specifically to improve sexual, reproductive and maternal health for women and girls, and to support more effective and equitable preparedness and response to epidemics and pandemics. When I say contextualized and responsible AI, what I mean is AI or AI innovations that are ethical, that respect human rights, that are inclusive and equitable, and that contribute to environmental sustainability. So that's our AI4GH initiative in a nutshell.
- [23:50]
A
Amazing. Great to hear that. And as a Canadian myself, I'm always thrilled to hear about Canada taking the lead on the global stage, particularly in an area like this. Sam, as I mentioned at the start, IDRC is one of the first organizations to fund artificial intelligence for low and middle income countries. Can you give us a peek behind the hood? Can you tell us a bit about the backstory? Like how did this initiative come to
- [24:16]
C
be now that AI is all the hype? And even before then, just before it became sort of on everyone's lips in the general sense, we recognized the potential that AI had to contribute to the achievement of the Sustainable Development Goals through, for example, catalyzing new startups that could work on various issues, whether it's health care, whether it's improving our food system, whether it's tackling climate change or even education. We saw that potential, but we also saw risks. This also draws on our experience of working in low and middle income context. We saw that there are risks that could generally be overlooked in the global north where things tend to work a bit differently. For example, we saw that AI could potentially reinforce some of the structural inequalities and biases that we have in these contexts. It could perpetuate gender imbalances, threaten jobs in contexts where jobs are already very hard to come by, but even it could facilitate more sinister acts, for example oppressive government surveillance in authoritarian contexts. So I think all this ultimately led to the establishment of what we are calling AI4D or artificial intelligence for Development program, which initially focused on Africa, was co funded by the Swedish funders Swedish Cedar, but is now growing beyond Africa and has brought on board other major funders like UK, the UK's FCDO and the AI for GH initiative that I mentioned earlier in a sense is an offshoot or perhaps a younger sibling of our wider AI 4D programming. Ed, I think the main call to action that I would have for funders who probably want to come into this space is that they need to be intentional about empowering low and Middle income country actors to take their destinies into their own hands. Whether it's creating space for AI innovators in the Global south to thrive and to scale, whether it's strengthening the policy and regulatory environment, whether it's helping the construction and collection and building and management of unbiased and equitable data sets, we know that AI needs Big Data. We need some more intentionality. And we simply can't rely on the goodwill of Big Tech and not trying to say anything negative about Big Tech, but we all know that Big Tech has been driving the AI revolution across the world, but their motives are primarily profit driven, as it should be. So I have no issue with that whatsoever. But to unlock the full potential of AI to do good in a more equitable way, then I think that those of us in international development, we need to play catch up, we need to play our part and we need to be intentional.
- [27:42]
A
Sam makes a couple of interesting points there, particularly about the unique role of philanthropy, of research, of funding agencies to combat, direct and shape the natural market forces of Big Tech and the private sector. We all work within an ecosystem and it's important to recognize the role you play in that ecosystem. My next question for Sam, again on the AI for Global Health program is how does it work? How is it structured?
- [28:12]
C
Our strategy for global health focuses on two broad thematic areas, so on sexual, reproductive and maternal health, and on epidemic pandemic preparedness and response. Now, across these two issues, we've decided that what we needed to do was to set up innovation hubs, AI innovation hubs that addressed these two thematic areas. And these innovation hubs will be tasked with sort of incubating cohorts or networks of innovators who are working at the intersection of AI and either SRMH or epidemic pandemic preparedness and response. Now, on the SRMH side, the sexual, reproductive and maternal health side, what we did was to set up regional hubs in four of the regions where we work. In Middle east and North Africa, in Sub Saharan Africa, in Latin America and the Caribbean. In Asia, we have an AI for SI MH hub hosted at a reputable institution in those respective regions. In Africa, for example, the AI for srmhub is hosted at the Infectious Disease Institute, which is associated with the Makerere University in Kampala in Uganda. On the epidemic and pandemic preparedness side of things, we have a global hub which is actually hosted in Canada at York University. And that hub, it has a more global purview, but is also working in those four regions And I think as of now they've sort of through a competitive process, They've identified over 16 projects on AI and epidemic preparedness and response that they will be supporting going forward. Now, same approach with the regional hubs. They also have cohorts of what we call sub grantees that they are working with. And all of them have come up with different innovations. Whether it's a chatbot for educating young people, whatever it is, they have different AI for health innovations that they are working on. But these hubs are largely working with, I'll call them more, they have a more research mandate and we recognize that research is important to sort of sow the seeds and do some de risk kind of investing. But you still need to think about commercialization. You need to think that some consider that some of these innovations might actually have a realistic pathway to become commercial goods, commercial products that could serve their respective contexts. And so for that we set up what we're calling AI commercialization hubs. So separate from these more research focused hubs. So there are three of them and each of these hubs are located, there's one in Africa, which is hosted at Vilgro Africa and is the same model that's replicated in Asia and in Latin America and the Caribbean where you have this institution that is known for being an incubator, an accelerator. We've given them funds and they're identifying again a cohort of startups that are working on AI for health innovations and working with them to incubate them, connect them with funding and put them on the pathway to commercialization. Then holding all that together, we have two other partners, the Global Health Network, health AI that are supporting all these regional hubs, the global hub, the commercialization hub, ensuring that whatever they're doing embodies inclusive AI, is responsible and brings all those things together to ensure that we are all speaking the same language, so to say, and playing from the same, the same playbook and ensuring that we don't replicate some of the things we are concerned about when we talk about, you know, biases and ethical concerns. So we are not looking at a top down approach. We are looking at innovations that are locally derived, locally led and locally driven. So that's what really excites us about this approach.
- [32:41]
A
Fascinating. And I'll definitely have to knock on some doors next time I'm in Nairobi or in Bharara. Indeed, of all the different research programs and the startups that you're supporting through this ecosystem, do you want to just give us a teaser of one as an example like one specific program or product or startup that is working in the AI space just to make it tangible.
- [33:03]
C
One of the ones that really excites me is again looking at how to educate young people using an AI chatbot to educate them about srhr. And it looks like something that is like a no brainer. Right. And possibly not even novel. But when you think about the fact that in many contexts young people have their own language, I don't know how it is in Canada or I don't speak it anymore. Well, it's the same here. Right.
- [33:32]
A
It is a different language.
- [33:33]
C
It is a different language. Right. For example, here in Kenya. So in Kenya there are two official languages. There's English and there's Swahili, but the young people speak Sheng. And Sheng is really a combination of Swahili and English. And it doesn't matter how many years you spent in school learning English and Swahili, you'll never get much of what they're saying. Now imagine trying to program a chatbot to be able to converse in Sheng. That is not a small feat. Yeah, that's the language of the young people and that's the language they're comfortable conversing in. And we know all the stigma around srhr. So you need to get it right.
- [34:20]
B
Right.
- [34:20]
C
And so, so that is. Those are some of the kinds of things that, that these groups are working on.
- [34:27]
A
That's awesome. That makes sense. Did you see some of the ideas, proposals, pitches that came through that didn't get funded? And could you talk a bit about what is not fundable from the perspective of IDRC or its partners?
- [34:42]
C
Yeah, I don't want to make people feel bad, so I don't want to
- [34:46]
A
be no name names.
- [34:47]
C
The applicants are bad. So I'm not going to go into specific, but just allow me to speak more philosophically about what we don't fund. And I think one irritating assumption that we've seen in other contexts and possibly some of the well intentioned applications that we get, is that people think they can copy paste ideas from one context to the next. And this tends to happen especially UN people think they can copy paste from the global north to the global South. It's well intentioned, it makes a lot of sense. If this AI application worked well here, it'll probably work well there. It makes sense. Right. But we've seen that assumption unravel and fail time and time again. And so that's why as idrc, we don't believe in what people call parachute research. So not research about parachutes but when people fly in with, with this great idea, they test it and they fly off and go and publish. We don't fund that kind of research. We don't support that kind of research. For us, the research needs to be locally led, locally driven, needs to respond to the context. And that's not to say you can't adopt an idea from elsewhere. You can, but you need to be very, very mindful about how that idea is developed and how it's rolled out. And you have to have the people on the ground involved in that process. So allow me to just leave it at that high level that that's the kind of work we'll just not support.
- [36:27]
A
Generally speaking, we will regretfully leave it there. All right, Sam, for our audience who's listening, we have many people joining from various different kinds of social impact organizations. Are there any specific resources on AI that you'd like to recommend for our audience?
- [36:44]
C
Absolutely. I think for anyone who is either interested in or working at this intersection of AI and health, there is this recent technical brief that was published by who and it's called Ethics and Governance of Artificial Intelligence for Health. I think it's an excellent document and they provide guidance on large, what they call large multiple multimodal models. Right. LMMs. But I think the whole, the principles that guidance sort of espouses and tries to promote, I think that should be almost required reading for anyone in the AI and health space. And they have all the other specific technical briefing documents as well. There's one on AI and sexual reproductive health, and those are also really good reads. So I highly recommend, recommend taking a look at those documents.
- [37:48]
A
Fantastic. Sam, I know in addition to your day job at the idrc, during nights and weekends, you also play a phenomenal role as the host of the MedXTech Africa podcast, which interviews health tech founders across Africa. Can I ask you to give a shout out to one or two startups that you think are doing awesome work in the space of AI for health? Speaking for yourself, not for idrc.
- [38:11]
C
Oh my. You're putting me on the spot. And I don't like to. I see them like, as my, like my children. I don't want to play favorites, but allow me to.
- [38:20]
A
You only have time for one or two?
- [38:21]
C
Yeah. Allow me to put them into two broad categories. So there are startups that are using AI to advance medical diagnostics and imaging. So big shout out to the likes of MenoHealth, AI Labs and Intexel. They're both using AI to advance radiology services. And then the other category startups that are using AI to improve patient engagement and access to treatment. So shout out to the likes of Huron AI that is making cancer care more accessible and Jacaranda Health that is empowering pregnant women in low resource settings to seek care when they need it.
- [39:00]
A
Awesome. What is one innovation or change that might make a big difference to address the global health challenge in the next 10 years?
- [39:11]
C
I'm particularly excited, and maybe this might sound like a cliche, but I'm particularly excited about the potential of AI assistance powered by regenerative AI. Their potential in taking healthcare closer to underserved communities. For example, there's this new device that is receiving so much attention, albeit for the wrong reasons. I don't know if you know about it, it's called the Humane AI pin. Now it's a small device that you pin or clip to your pocket or to the lapel of your suit or whatever and it helps you to interact with AI almost as if it were a person. So you could be walking down the street, you see a stunning plant that you've never seen before and you'd like it in your garden, but you don't know what it's called. So you ask Humane AI to take a look and tell you what plant that is and it tells you right there and then. So it's almost like having a pocket friend at your fingertips. But unfortunately it doesn't work very well and all the tech reviewers have basically called it trash. But the idea is to. Yeah, sadly. But the idea is still quite fascinating. I can imagine a community health worker in a remote community using that device to detect, say, let's say there's a brand new outbreak of a rash that no one knows about in that community. And you know that community health worker can use that device and it could tell it what it is, it could tell the community health worker what that rash is, give that community health worker real time guidance on how to handle it, and even trigger any reporting that needs to be done to the health authority. No need to google, no need to take pictures, no need to go through some complex clinical decision tree process. You just have this AI device that is your companion, your assistant, and it helps you be more effective in serving local communities in areas that would ordinarily not be able to get access to such advanced high quality care. So that's what excites me. Maybe I'm naive and maybe we are many years away from that, but such things really, really excite me.
- [41:30]
A
That sounds awesome. I want one even just to take care of my kids. That sounds amazing.
- [41:35]
C
As Dr.
- [41:41]
A
Well, guys, that's it. That's our show for today. Big shout out again to our sponsors, Reach Digital Health for making this podcast miniseries possible. And of course, all of you, our listeners, for joining us today. Stay tuned for the next and final episode in this miniseries where we dive under the hood of AI for health speak with the implementers and data scientists actually building out novel AI systems for health in Africa. To get notified when these episodes air, don't forget to subscribe to this podcast. And if you want to stay in the know about what's going on with healthcare ventures in Africa, you can also sign up to our newsletter@africahealthventures.com newsletter. Have a great rest of your day. I'll see you soon.