- 18 hours ago
"As AI becomes embedded in healthcare, from data analysis to care delivery, it is reshaping how decisions are made. But who is represented in the data, who benefits from these systems, and who risks being left behind?
Bringing together perspectives from policy, technology, and healthcare, this session explores how to build truly equitable AI-driven systems. How do we address bias in data and algorithms? What role should public institutions, companies, and civil society play? And how do we move from principles of equity to measurable impact, ensuring AI strengthens, rather than fragments, access to care?"
Bringing together perspectives from policy, technology, and healthcare, this session explores how to build truly equitable AI-driven systems. How do we address bias in data and algorithms? What role should public institutions, companies, and civil society play? And how do we move from principles of equity to measurable impact, ensuring AI strengthens, rather than fragments, access to care?"
Category
🤖
TechTranscript
00:18Hello, everyone. Welcome, welcome.
00:22So, my name is Mujan Asghari.
00:25I am very delighted to moderate this panel with our beautiful, amazing panelists today.
00:32You're going to have a topic that probably is something you've been hearing a lot about equitable systems.
00:40This time it's about healthcare.
00:42So, let me tell you a little bit about myself and the panelists, and we're going to dive deeper into
00:46this topic.
00:47I am the co-founder of Women in AI, and Women in AI is a nonprofit organization across 150 countries,
00:54purely to support women getting access to education and to these technologies to be part of it.
01:00So, they're not just using these technologies, but they are the center creators of this technology.
01:07I'm also the founder and CEO of Thousand Faces, which is a funding platform for female founders building impact startups,
01:15including healthcare.
01:16And I'm very delighted, excuse me, today to have with us four amazing panelists.
01:27So, introducing first, Dr. Joanna Hackett.
01:31Joanna, you're the VP and Global Head of Health System Services at IQVIA,
01:37working with governments, health systems worldwide at the intersection of data, genomics, and transformation.
01:46Welcome.
01:51We have Dr. Alexander Ni, the president of Tencent Healthcare,
01:58and the person who leads the Inclusive Health Lab, a resource-limited setting across Asia.
02:05He's also an advisory to the WHO Digital Health Technical Advisory Group.
02:12We have Anca del Rio, co-founder and executive partner at Acuvera,
02:20and WHO consultant, where she works with health system leaders, policy makers, industry, and investors
02:28to turn AI into system-level impact.
02:31So, welcome, everyone.
02:33And the last person, Juliette Moreau.
02:36She's a co-founder of Femtech Friends, and founder and CEO of My S Life,
02:43building the women's health innovation ecosystem across Europe,
02:47and advocating for digital solutions that are designed around women.
02:53Thank you so much, everyone, to be here.
02:55And let me tell the audience today a little bit about the topic.
02:59So, we talk a lot about what AI can do for healthcare, the breakthroughs, the efficiencies, the promises of personalized
03:10medicine,
03:10but there's a question that doesn't get asked quite enough.
03:15Who is this actually for?
03:17Because right now, somebody can wait for 18 months for an MRI in a public hospital,
03:25where just in the same street next door, there are empty machines for a private healthcare system.
03:33Women are still excluded from clinical trials because of their biology that doesn't fit the standard model.
03:41And the question is, where did this standard come from?
03:46So, the entire healthcare system, not just individuals, risk being left behind
03:52because they simply don't have the infrastructure, the governance, the financing to absorb AI at the speed that is moving.
04:04So, I would like to start the first question.
04:07What does an equitable healthcare system actually look like?
04:11So, I want to start with you, Joanna.
04:13You've spoken about the gap between public and private systems, the MRI queues, the case for prevention over treatments.
04:23What does fixing that system actually require before AI even enters into the picture?
04:31Great.
04:32So, that question we could probably spend the next four to five hours discussing.
04:37However, I will distill it down to a couple of different points.
04:41It is true that there is inequity, and this is not just in technology.
04:45It's not just in gender.
04:47It's not just in location.
04:49And the reason for this inequity, generally speaking, usually comes down to the fact that
04:54we don't prioritize thinking about prevention over treatment.
04:58And if we thought more about a preventative healthcare system, we wouldn't have to necessarily think about the waiting,
05:06the queues, the trying to get technology embedded into healthcare and to doing things a bit differently.
05:11So, I'm not saying that if we all, you know, ate better and exercised more, that we would be in
05:17a much better situation.
05:18But it does come down to a little bit between personal choices as well as the incentives in the system
05:25to make sure that we are rewarded with the quality of healthcare that we actually need.
05:30So, distilling it down to basic principles, to try to make sure that there's more equity in healthcare as a
05:36whole,
05:36we have to make sure that there is a link between a personal incentive to take better responsibility for our
05:43healthcare
05:44and also a societal imperative to make sure that we're actually connecting the two together.
05:49We can't expect ourselves to just be sitting around waiting for someone to give us drugs to feel better
05:58at the same time that we're not necessarily willing to take preventative measures, to take the stairs and things like
06:03that as well.
06:04So, it's not all about prevention, but a lot of the inequity, and I'm not talking about, you know, areas
06:11of the world
06:12where we don't even have very basic healthcare, it's general principles as a whole.
06:17And this is where a lot of governments are thinking about that preventative and longevity agenda much more
06:23before we're even thinking about the treatment.
06:26Thank you so much, Johan.
06:28Alex, you've built Tense and Inclusive Health Lab, specifically around this question.
06:35From your experience across China and resource-limited settings,
06:39what does the inequality gap actually look like on the ground,
06:44and where do you see the biggest structural opportunity for AI?
06:48Thanks, Mojang, for the question.
06:51In your stunning remarks, when you talk about inequality,
06:55you mentioned both, I guess, waiting time and potentially cost.
07:01But I think when it comes to inequality in healthcare,
07:05you always need to look at three dimensions.
07:07One is access, second is cost, and then the third one, I think,
07:11even actually quite important is actually quality.
07:14And, I mean, I worked around the world,
07:16and I think the way each country and each setting manifests inequality very, very differently.
07:21Now, I think in the case of China, cost is actually not a big issue
07:26because we have got a single-payer social insurance system,
07:29and the cost of these investigations tend to be quite cheap.
07:34So, if cost is not an issue,
07:36access, there are actually enough hospitals and clinics around.
07:39Now, then, the last one is then quality.
07:42Quality is actually quite disparate between the large academic centers
07:47versus the rural settings, right?
07:49I think that also applies to a lot of countries as well.
07:51And so what we're trying to do with the Inclusive Health Lab
07:54is that, well, let's look at the issue fundamentally.
07:59Is it how can we improve the quality in the most low-resource setting?
08:03And I think this is where AI can be very, very powerful.
08:07And at the same time, access, we think about traditional access I mentioned.
08:12China has a lot of hospitals and clinics around,
08:14but I think increasingly people are also looking at access
08:17in terms of digital access inequality.
08:20Now, not that China has a digital access issue
08:23because everyone has a smartphone,
08:25and 5G coverage is pretty much ubiquitous.
08:28But then the issue is about digital literacy.
08:31There is actually a lot of elderly, right,
08:33who is not up to this whole wave
08:35in terms of adopting the digital access.
08:38So then we now actually, the last 10 years,
08:41we've incubated a new problem
08:42where you have another marginalized group, right,
08:45that could not access to healthcare
08:46if they're not fluent in the digital platform.
08:49And so what we're trying to do
08:51is look at this as a critical issue.
08:54And one of the reasons why we focus on survival cancer,
08:57because it's actually a huge problem still,
09:00from awareness to access to screening,
09:04then to treatment.
09:05And I think along this whole journey,
09:07we try to basically embed better AI
09:09to improve the quality in terms of education
09:12and messaging and diagnosis,
09:13but at the same time on how to make sure
09:16that the cost is very, very low
09:17so that we can actually reach as many people as possible.
09:21Thank you so much.
09:23Juliette, through Femtech France,
09:26you are representing a population
09:28that is literally half of humanity
09:30and remains underrepresented in research,
09:33diagnosis, and in innovation.
09:37What does, in your point of view,
09:39an equitable healthcare looks like,
09:41especially for women?
09:42And why do you think it has taken so much time
09:45to reach to that equality?
09:48Well, obviously, an equitable system,
09:52health system, will include women
09:54the same way they include men.
09:57And thank you for pointing that.
09:59The issue we are facing today
10:00is that although women live longer,
10:02and I think we all know that by now,
10:04the years of life they will spend in good health
10:09are less compared to men.
10:12And that's the main issue.
10:15You shared it.
10:17They have difficulties being included in research.
10:21There are misdiagnoses.
10:24They are not listened to when they have some symptoms.
10:27And we could claim a lot of reason for that,
10:32the first one being potentially patriarchal,
10:34but I won't go that way.
10:36I think that we have, for a lot of years,
10:41understood equality as sameness,
10:45and that's where we need to pay attention.
10:49That's where we need to change the way
10:52we are looking at women,
10:54at health and research in health today.
10:56And this is leading to the necessity of regarding women,
11:02and then we can go in a lot of different sensation after that,
11:06but women as a body that has different needs
11:10in terms of health than men,
11:12and from that on,
11:15it's a way of trying to build a more equitable health care.
11:20And that's what we are doing with Femtech.
11:21We are trying to build innovation
11:24that will reduce the gap
11:27between women's health and men's health
11:30by looking at both specific conditions to women,
11:34and I could quote endometriosis, for example,
11:37but also looking at conditions
11:39that are affecting men and women,
11:41but where we know there are different symptoms,
11:44different reactions to treatment,
11:46and potentially different ways
11:47of taking care of the people that are affected.
11:53Anka, I want to come to you.
11:56We talked about it that basically
11:58the gap often starts at the individual level
12:02with who is represented in the data
12:05and who can access the care at all.
12:09And your efforts at your company, Acuvera,
12:13is particularly around health care delivery design
12:16and health tech investments.
12:19And you're pointing out to a second layer,
12:21which is whether the health system
12:24has the incentives, enough incentives,
12:27to actually do something about it
12:29and try to deliver on the AI promises.
12:32Can you...
12:33Where do you see this conversation to start?
12:35Is it on the patient side
12:37or on the system side?
12:39Thank you, Mujan, very much for the question.
12:43I would say the data problem is solvable.
12:47But when we look at the delivery problem,
12:50that's structural.
12:52And unfortunately, it propagates
12:54across many health care systems these days.
12:57And I would say that, yes,
13:01the problem starts with who's represented in the data
13:05and who can access care at all.
13:10But what we see at Acuvera is the layer underneath that.
13:15It's basically what health care systems
13:19have the capacity and capability
13:22to actually deliver on what the data shows
13:26and what the AI promises.
13:28And by that, I mean the workflows,
13:31the workforce,
13:33and the incentives that Joanne briefly mentioned.
13:38When we look at who's financing AI
13:41and how can we ensure that that is financed sustainably
13:47and not just widening the gap
13:49in accessing what AI can do.
13:53And by saying that,
13:56I would like to just swing back one bit
13:58on the points that Juliet made
14:01because I couldn't help but noticing
14:04the room is filled with female audience,
14:10which is applauding, of course.
14:12But as a female speaker and working so much
14:17and putting a lot of effort
14:19across the European Commission,
14:22UN agencies,
14:23and now sliding into the private sector
14:26trying to change things from within,
14:29it hurts.
14:30Because I would like to see
14:33how many women
14:35are actually on the stage,
14:38on the governance,
14:39in the government stage.
14:41And I would like to see more men
14:44in this room,
14:45applauding the ones who joined us,
14:48and see how...
14:53And seeing how the decision-making layer
14:56is actually shifting.
14:59We're going to get to the decision-making
15:02and governance question very soon.
15:04What I'm hearing now is
15:06actually that the problem is much deeper.
15:08That's what we're agreeing on.
15:10It's structural.
15:11It's cultural.
15:12And it's about who gets counted
15:16and who doesn't.
15:18And even before a single line of code is written.
15:24And there's a question.
15:26We had it in the proportional call
15:28that if the system,
15:31and it sounds like it is the case,
15:33if the system is already fragmented and broken,
15:35how can we even build on top of that?
15:39Something happened during the pandemic
15:41that we had to accelerate
15:43this digitalization
15:46and developing system
15:48that could help us in the midst of crisis.
15:52And many of these accelerations
15:54for the technology
15:55has been done
15:56without that much of a care and review
16:00of the human.
16:01And now after the pandemic,
16:03post-pandemic phase,
16:04we are shifting our attention
16:07towards where now new priorities are
16:10and sort of passing
16:14by what it has not been well done
16:17and well created during those phases.
16:21So the question would be,
16:24what is the point of building AI systems
16:26on top of what is already fragmented and broken?
16:30I want to come back to you, Joanne.
16:33You had this very clear point
16:35about an agent.
16:39If the underlying data is broken,
16:42what can we do about it?
16:44What can we do
16:45if the underlying data is already flawed?
16:48How can we go about that?
16:50And you gave an example
16:52about women being excluded
16:55from clinical trials
16:57because of the way their biology is,
17:00because of their menstrual cycle,
17:01a cycle that can disrupt the test,
17:07the clinical trials.
17:09What needs to happen
17:11in the data and the infrastructure level
17:15before the AI is actually built
17:17and even being trusted
17:18by clinical settings?
17:23This is often one of the things
17:25that we end up grappling with
17:27because we can say
17:29that all data should be digitized.
17:31It should be everything is connected,
17:34ambulance records, GP records,
17:36all of this kind of stuff.
17:37And I'm not saying
17:39that we shouldn't aspire to that,
17:40but the reality is
17:41that's not where we are today.
17:43That doesn't mean
17:44that we can't be innovative.
17:45Now, you mentioned the pandemic.
17:48That was when people did
17:50some pretty interesting things
17:51in healthcare
17:52and we were able to achieve
17:54some pretty remarkable outcomes.
17:56It's not that all rules
17:57were thrown out of the window
17:59and we just all went a bit crazy.
18:01It was because,
18:02if you think back to it,
18:04most of us,
18:05I'm assuming every single one of us
18:07were affected greatly by this.
18:09What was the one thing
18:10almost every single person did?
18:13They were willing to trade information
18:15for outcomes.
18:17All of us, you know,
18:18carried around the QR code
18:20to either go to a restaurant
18:21or to leave our homes
18:23or to exercise
18:23or to go and get a vaccine.
18:25Every single person
18:26could understand
18:27the value of that trade-off.
18:29And this is where sometimes
18:31the underlying data principles
18:33are flawed
18:34because people don't understand
18:36where that trade-off
18:38is going to be.
18:39So it's great to say that,
18:41yes, if all GP data
18:43was federated,
18:45standardized, cleaned,
18:47well, before you even think
18:48about doing that,
18:49where are you logging
18:51this information?
18:52Is it being shared?
18:53And if the answer is no,
18:55why?
18:56And very often,
18:57we don't see the value
18:58of doing that.
18:59So if we don't see
19:00the value of doing that,
19:02most people either don't care,
19:04say no as a default,
19:05or don't want to necessarily,
19:07not very motivated
19:08to change it.
19:09So I think that we really
19:11have to change the thinking
19:12as to how do we engage citizens?
19:15Let's not wait until we're sick.
19:17How do we engage citizens
19:18to think more proactively
19:20about that whole healthcare continuum
19:22from zero to 150?
19:25Because probably many of you,
19:26many of you young people
19:27in the room
19:28will be 150 at some point.
19:30How do we think about that
19:31and understanding
19:32how the subtle nuances,
19:34how can we pick up the patterns,
19:35how can we diagnose someone
19:37who's nine years old
19:38who's going to get
19:39early onset dementia
19:40when they're 60?
19:41And I don't think that we're,
19:42we don't spend enough time
19:43thinking about living
19:45good quality lives.
19:47We think more about
19:48how do I get treated
19:49when I'm sick?
19:50So to me,
19:51the underlying principles,
19:52and yes,
19:53I would love to,
19:54you know,
19:54do OMOP standardization
19:56for data
19:57and all of these
19:57other fun things,
19:58but the data needs
19:59to exist first
20:00in order for us to do it.
20:01So to me,
20:02some of those
20:03building the AI layers
20:04on top of that
20:06is,
20:06that's the easy part.
20:08We need to collect the data
20:09and make people understand
20:10why we're doing it.
20:12Your point about
20:13different clinical trials,
20:15it's not that women
20:16are excluded from them.
20:18We've not built data patterns
20:19to understand
20:20if the eye pressure
20:22is normally a little bit higher
20:23during a menstrual cycle.
20:24We need to bring that
20:26into consideration,
20:27not excluded completely.
20:29So to me,
20:29it's understanding
20:31what we're trying to achieve
20:32and gathering the right
20:34data points
20:35and principles
20:35and making sure
20:36that that relationship
20:37about the understanding
20:38why this is happening
20:40is so much more important.
20:42The rest of it's easy.
20:44Thank you, Joan.
20:46Alex,
20:47so we talked a little bit
20:49about the cervical cancer screening
20:51that is part of your work
20:53in rural China,
20:54which is actually exactly
20:55talking about this tension.
20:59When traditional methods
21:01heat cultural
21:03and access barriers,
21:05your team actually has
21:06had a completely
21:08different approach to that.
21:09So could you talk
21:10a little bit about
21:10what have you done,
21:11what have you seen?
21:13We talked a little bit
21:14about the moving
21:15to the urine
21:17and blood tests
21:18that you found a way
21:21to go around it
21:22when there is not
21:24enough data
21:25and being able
21:26to bypass those barriers.
21:27Could you a little bit
21:28talk about that?
21:30Thanks, Mujan.
21:31I think it feels strange
21:34for me just to go through
21:35with a lot of female
21:36on board, right?
21:37What you need to go through
21:38for cervical cancer screening,
21:40right?
21:40The old traditional
21:41clinic type pap smear
21:43and then you do follow-ups
21:44after that.
21:45But this model,
21:47when we were actually
21:48looking at the issue
21:50of cervical cancer screening
21:51rate in China,
21:52the issue is,
21:54well,
21:54when people look at the issue,
21:56screening rate is low.
21:57Then what do you do?
21:58The traditional method is
21:59we need to train more people.
22:01We need to have more clinics.
22:02We need to tell more women
22:04to come.
22:04But the issue is
22:06that is actually
22:07not sustainable, right?
22:08And not just economically,
22:10but just time frame-wise.
22:12It doesn't make sense.
22:13Then we start looking at it.
22:14Well, what are we trying to...
22:15Basically,
22:16we try to triage
22:17who needs to be there.
22:18And then this is where
22:20we got the scientists
22:21on board and say,
22:22look,
22:22we need to think about
22:24it very, very differently
22:25on how we look
22:27at cervical cancer screening.
22:28The way that we tackle
22:30other forms of cancer screening
22:31before is look at the root cause.
22:33The root cause is basically
22:35largely by itself HPV.
22:37But vaccination is not available
22:39in China at that time,
22:40especially for a lot
22:41of the elderly women.
22:42Now,
22:43then what we ended up doing
22:45is that we have
22:46a very, very interesting way
22:47of being able to detect
22:50signals of HPV infection
22:53or past infection
22:54as well as cancer changes
22:56just from urine.
22:58So that also helped us
23:00inadvertently solve
23:02another issue
23:03in resource-limited setting.
23:05In rural China,
23:07there's a big population
23:09of ethnic minorities, right?
23:11Not different from Han Chinese.
23:13And for their own
23:15ethnic minority culture, right?
23:17Basically,
23:17it's very conservative.
23:19Even with a female doctor,
23:21they would not let you
23:22do a pap smear, right?
23:23And I think this
23:24urine-based sampling, right,
23:26actually addresses
23:27multiple issues.
23:28But I think
23:28that requires a lot of the times
23:30just thinking out of the box
23:32rather than
23:32just trying to expand
23:34what you do differently, right?
23:35But just going
23:36to the first principle.
23:37All I want
23:38is to do risk stratification.
23:40And how do we actually
23:41look at this problem
23:42very differently?
23:43And I think this is
23:44where we came up with this idea.
23:45We're still doing it,
23:46doing clinical trial, right,
23:48to prove then
23:48the ultimate effectiveness.
23:51And ultimately,
23:52something that we develop
23:53for resource-limited setting,
23:55we have a dream, right?
23:56It might actually become
23:57the new standard, right?
23:59Because I don't think
24:00anybody enjoy going
24:01to the clinic
24:02on a regular basis
24:03to do it.
24:04If you can do it at home
24:05with a urine sample,
24:06I think a lot of people
24:07will choose to do it.
24:08But then,
24:09if you just think about it
24:10from a resource-rich setting,
24:12you'll never come up
24:12with a solution.
24:14Very interesting.
24:16Thank you, Alex.
24:17Anka,
24:19we basically
24:21talked a little bit
24:22about your work
24:25at WHO
24:26and the article
24:28that you wrote
24:29about Eugenic AI
24:30and the risk
24:31of building AI-powered
24:32health systems
24:33that they don't have
24:34the governance
24:34and cybersecurity readiness,
24:37financial sustainability,
24:38and so on.
24:40I would like to know
24:41what do you think
24:43does require
24:44to create
24:45a resource foundation
24:47to be able
24:48to create a system
24:50where it actually
24:51can work
24:54in a way
24:54that we need it
24:55in an equitable way
24:56and who can pay
24:58for that?
25:00Thank you
25:01for the question,
25:01Mujan.
25:03Let me start
25:04by saying
25:04that only 4
25:07out of 50
25:08WHO
25:10European countries
25:12have an AI strategy
25:15in place
25:16particularly made
25:18for health.
25:194 out of 50.
25:21And you can let
25:22that sink in.
25:23And then
25:24when we dive
25:25one layer deeper,
25:26we see that
25:27only 4,
25:29again,
25:30have a liability
25:30framework in place
25:32for when AI
25:34goes wrong
25:35and causes harm.
25:39This is
25:40the extremely
25:41dangerous layer
25:42that we tend
25:43to overlook.
25:44So,
25:45back to your question,
25:46when we ask
25:47what a properly
25:48resourced
25:48foundation requires,
25:51I would say
25:51that almost nobody
25:53has built it yet.
25:56I think
25:57until financing
25:58and liability
26:00are not built
26:02within
26:02as infrastructure
26:04in health systems,
26:06not as an
26:07afterthought,
26:09whoever pays
26:11for AI
26:12and the gap
26:14that it's producing
26:15is the one
26:16least capable
26:17of absorbing
26:18the cost.
26:20So,
26:21what we are
26:23trying to do
26:24and putting
26:25a lot of effort
26:25into is working
26:26with not just
26:28healthcare delivery
26:29settings,
26:29but also
26:30with payers
26:31and looking
26:32at their
26:35payment models
26:37and how
26:38that healthcare
26:40delivery
26:40shifts
26:41together
26:42with AI
26:43across the
26:46whole system
26:46and not
26:47in the gaps
26:48that,
26:50unfortunately,
26:51it risks
26:52widening.
26:57Juliette?
26:59many areas
27:00of women's
27:01health
27:01is still
27:02suffering
27:02from a
27:03fundamental
27:04lack of
27:04investment.
27:05So,
27:05talking about
27:05the investment
27:06perspective
27:07again,
27:09can AI,
27:10in your opinion,
27:11close that
27:12innovation gap?
27:13Or is there
27:15a risk
27:15that it
27:16simply
27:17concentrates
27:18the investment
27:19where it has
27:20been already
27:20focusing
27:21and in a way
27:23to widen
27:24the gap?
27:24What do
27:25you think?
27:27I'll try
27:28to be
27:28optimistic.
27:29So,
27:30I'll start
27:31with the
27:31good news.
27:33You spoke
27:34a lot about
27:34the difficulties
27:35that we have
27:36to include
27:36women in
27:37clinical research
27:38and clinical
27:39trial.
27:39One of
27:41the reasons
27:41is that
27:42women are
27:43a risk
27:43when you
27:44do clinical
27:45trials
27:46because they
27:47may carry
27:48a baby
27:48and this
27:49baby may
27:50suffer
27:50from the
27:51treatment
27:52that you
27:53are testing.
27:54The second
27:55risk that
27:56women represent
27:57when they
27:57get older
27:58so that
27:59when they
28:00pass
28:00menopause
28:00and they
28:01can't
28:01have
28:01baby
28:02anymore
28:02is that
28:03they
28:03are
28:03alone.
28:04Usually
28:05they survive
28:05their partner
28:06and you
28:07need the
28:08partner
28:08to come
28:08to bring
28:09you to
28:10the clinical
28:10trial
28:11and take
28:11you back
28:12home
28:13especially
28:13for example
28:14when you
28:14are testing
28:15people on
28:16Alzheimer
28:16and this
28:18is also
28:18another reason
28:19why women
28:19are not
28:20included
28:20in clinical
28:21trial.
28:22They cost
28:22more.
28:23Good news
28:24with AI
28:24and we
28:25see some
28:26companies
28:26developing
28:27that right
28:27now
28:28is that
28:28you can
28:29build some
28:30model that
28:31will help
28:31you work
28:32on very
28:32smaller
28:33cohort
28:34and help
28:35you to
28:35have a
28:35much more
28:37heterogeneous
28:39population
28:42on which
28:42you are
28:43going to
28:43test
28:44and do
28:45your
28:46clinical
28:46trials.
28:46That I
28:47would say
28:47is good
28:48news
28:48because
28:48on the
28:49investment
28:49perspective
28:50you will
28:51be able
28:51to pay
28:52less
28:52to have
28:53a much
28:54better
28:54view
28:54of how
28:55your
28:55treatment
28:56is
28:56working
28:56and
28:57to have
28:58also
28:59a better
28:59pathology
29:00so the
29:00capacity
29:01to know
29:02exactly
29:02what
29:03level of
29:04treatment
29:04you need
29:05to give
29:05to each
29:06type
29:06of
29:06people
29:07including
29:07based
29:08on the
29:09gender
29:09of the
29:09person
29:10so that's
29:11the good
29:11news
29:11the bad
29:12news
29:13is
29:13we can
29:14build a
29:15lot
29:15on health
29:16on AI
29:16and it's
29:17not at
29:17VivaTech
29:18that I'm
29:18going to
29:20prove that
29:21but the
29:23point is
29:23you need
29:24to enter
29:24data
29:25that are
29:25corresponding
29:26to the
29:26people
29:27that will
29:27be using
29:28that AI
29:29and the
29:32previous panel
29:33I don't know
29:34if some of
29:34you were
29:35hearing it
29:36was on
29:36longevity
29:38if you're
29:39building a
29:39model that
29:40will predict
29:40everything you
29:41need to do
29:42to live as
29:43longer and
29:44in the best
29:45health as
29:45possible and
29:46you miss
29:47menopause
29:48you lose
29:49half of
29:50your population
29:51and that's
29:52the main
29:52point if
29:53we don't
29:53include some
29:55data that
29:55are very
29:56specific to
29:57women and
29:57sorry to
29:58well I hope
29:59that everybody
29:59here will
30:00go until
30:00menopause
30:01that will
30:01mean that
30:02at least
30:02you've lived
30:03longer enough
30:04to experience
30:05that beautiful
30:06experience
30:06you will
30:07miss
30:09half of
30:10the population
30:10because you
30:11didn't enter
30:12the parameters
30:12that are
30:13necessary for
30:13that and
30:14that relates
30:15to investment
30:16because women's
30:18health is
30:18probably right
30:19now not
30:20invested enough
30:21it's where we
30:22miss most of
30:24the financing
30:25that we need
30:27and that's a
30:28shame because
30:29there is an
30:30economical potential
30:31in oneself that
30:32is enormous
30:32I hope that
30:33some investors
30:34are listening to
30:34that right now
30:36that's where
30:37you will probably
30:37miss some of
30:38the needs
30:40that are
30:41necessary to
30:41build an AI
30:42that includes
30:43women as well
30:45I want to
30:46continue with
30:47yourself for
30:48the next
30:48question because
30:49I think you're
30:49touching on a
30:50very important
30:52topic which
30:53is we're
30:53missing the
30:54data of
30:54women we're
30:55not including
30:56them what
30:58do you think
30:59is the main
30:59reason like
31:00first thing
31:00what do you
31:01think we're
31:01not doing
31:02that and
31:03what is also
31:03the incentive
31:05like do you
31:06think the
31:06systems today
31:07the governance
31:09that we have
31:10the organizations
31:12involved do
31:13do they have
31:14the market
31:14does it have
31:15the incentive
31:16to do that
31:16or it is
31:17actually not
31:19treated the
31:19data problem
31:20it is maybe
31:21the governance
31:22issue
31:25innovation follows
31:26incentives
31:26that's for sure
31:27and let me share
31:28a little story
31:29about one of
31:30our start-up
31:31the start-up
31:31within Femtech
31:32France
31:33great company
31:34the name is
31:35Dahlia
31:36the company is
31:37working on
31:38depression
31:39depression
31:40depression
31:40affects women
31:401.5 more
31:42than men
31:43but the company
31:44is curing
31:45everybody
31:46okay men
31:47and women
31:48as well
31:48they didn't
31:49they didn't
31:50get any
31:51difficulty to
31:51be reimbursed
31:52because obviously
31:53there was a
31:54path for them
31:55and they were
31:55working on a
31:56disease that
31:56was already
31:57funded
31:57when they
31:59started looking
32:00at postpartum
32:01depression
32:01and postpartum
32:03depression
32:03that's 10 to
32:0415 percent
32:05for the women
32:06that are
32:07so postpartum
32:08sorry
32:08it's the period
32:09that follows
32:10the end
32:10of the pregnancy
32:11and this period
32:13can lead
32:14for some women
32:1510 to 15 percent
32:16to a depression
32:18well they didn't
32:19find some funding
32:20to be reimbursed
32:21and I think
32:22this is it
32:23once you
32:24once you decide
32:25that you
32:25finance
32:26specifically
32:27some diseases
32:28that affect women
32:29that have an impact
32:31as well
32:31on the
32:32on the whole
32:33ecosystem
32:34of the family
32:35of course
32:35but all the
32:36also the
32:36economical
32:37ecosystem
32:38a woman
32:38that is
32:39in depression
32:41is someone
32:41that is not
32:42working
32:42that is not
32:43bringing money
32:44to her family
32:45as well as
32:45the society
32:46and is not
32:47paying as well
32:48for the healthcare
32:48system
32:49so this is a
32:50whole block
32:51if we decide
32:52that once
32:53and when I say
32:54we as
32:56sorry I will
32:57speak from a
32:58European perspective
32:59because that's
32:59probably the
33:01social protection
33:02system that I
33:02know the best
33:03but if we
33:03don't decide
33:04that tomorrow
33:05we will reimburse
33:07treatment
33:08we will reimburse
33:09healthcare innovation
33:11whether they are
33:12AI or not
33:13based on the
33:14capacity to
33:15give a service
33:17that is equal
33:18to men and
33:18women
33:19then we will
33:21miss the point
33:21and this is
33:23really how we
33:24are going to
33:24lead to some
33:25innovation to
33:27move to that
33:28place
33:28if not
33:28we will
33:29remain where
33:30we are safe
33:30men's health
33:35well let's
33:36talk a little
33:37bit about
33:37regulation
33:39we've been
33:39talking about
33:40it
33:40that the
33:42government
33:42the governments
33:43in different
33:44countries
33:44they have
33:45their rules
33:45some
33:46governments
33:46they don't
33:47even have
33:47a framework
33:47for AI
33:49the landscape
33:50is very
33:51fragmented
33:51and
33:54also the
33:54problem is
33:55that the
33:55technology is
33:56moving very
33:56fast and
33:57the regulation
33:58is lagging
33:59behind
33:59so how
34:00can we
34:01fill up
34:01this gap
34:02and who
34:03can
34:03who is
34:04going to
34:05be left
34:05behind
34:05if this
34:06gap is
34:07still
34:08existent
34:09so
34:09Joanne
34:11you work
34:12with different
34:12governments
34:13in different
34:15health system
34:17environments
34:17around the
34:19world
34:19what do you
34:20actually see
34:21when governments
34:22try to
34:23govern AI
34:24without the
34:26regulatory
34:26framework
34:26or
34:28the
34:28internal
34:29capacity
34:31very often
34:32if there's
34:33not a
34:33framework
34:34in place
34:35most of
34:36us
34:36regardless
34:36of what
34:37that
34:37framework
34:37should be
34:38is we
34:38become a
34:39little bit
34:39more
34:39conservative
34:40and it's
34:42not wrong
34:42to be
34:43conservative
34:43it's actually
34:44very helpful
34:45sometimes
34:45but if you
34:46don't have
34:47those guard
34:48rails
34:48and you
34:49don't
34:49understand
34:49what the
34:50art of
34:50the possible
34:51could be
34:52that default
34:53is always
34:53just to be
34:54same
34:55same
34:55let's not
34:56get a little
34:56bit too
34:57ambitious
34:57and that's
34:58where it
34:59would be
34:59a real
35:00shame
35:00if we
35:00don't
35:00end up
35:01having
35:01a very
35:02proactive
35:02framework
35:03for the
35:04governance
35:04side of
35:05things
35:05because we
35:06will then
35:06miss out
35:07on being
35:07able to
35:08adopt
35:09some therapies
35:10or adopt
35:11different ways
35:11of working
35:12think again
35:13going back
35:14to the
35:14COVID
35:15example
35:15as I said
35:16I'm not
35:17saying that
35:17that was
35:17the best
35:18time for
35:19innovation
35:20and health
35:20care
35:21but a lot
35:21of things
35:22were done
35:22differently
35:23because we
35:23could understand
35:24the steps
35:25that were
35:26probably in
35:27the way
35:27a little bit
35:28more so
35:28than we
35:29do today
35:29we've defaulted
35:31back to that
35:31pretty quickly
35:32because it
35:33wasn't put
35:34into legislation
35:34it wasn't put
35:35into law
35:36it wasn't put
35:36into policy
35:37so the easiest
35:38thing to do
35:39is again
35:39be a bit
35:40more conservative
35:41what we're
35:42seeing is a lot
35:43of governments
35:43now are trying
35:44to understand
35:45if we were to
35:45remove a lot
35:46of this barrier
35:47and I'm based
35:48in the UK
35:49if you're following
35:50anything that's
35:51happening in the UK
35:52healthcare system
35:53we've abolished
35:54the national
35:55healthcare service
35:56it sounds very
35:57grand
35:57we didn't get
35:58rid of our
35:58healthcare system
35:59we've gotten
35:59rid of a lot
36:00of the bureaucracy
36:01so there's
36:02things like that
36:03that is starting
36:03to happen
36:04because if you
36:05do look back
36:06on the way
36:07innovation has
36:08been created
36:08and rolled out
36:09we kept putting
36:10in more and more
36:11and more hurdles
36:12it wasn't to
36:14stop innovation
36:15it was to
36:15understand
36:16what's that risk
36:17how is it going
36:18to be governed
36:19and we've almost
36:19over regulated that
36:21so I think
36:22what we're seeing
36:22is that there's
36:23bureaucracy that's
36:24being taken out
36:25of many healthcare
36:26systems
36:26and possibly
36:28it was because
36:29of the pandemic
36:30and also many
36:31of the governments
36:32applied for
36:33this very special
36:35fund that was
36:36created called
36:36the EU
36:37Recovery and
36:38Resilience Fund
36:39and many of
36:40these governments
36:40they were putting
36:41together some
36:42very fantastic
36:43research initiatives
36:44to be able to
36:45understand how to
36:46do healthcare
36:47projects differently
36:48in their countries
36:48this is now
36:49five years old
36:50the outcomes
36:51of some of
36:52these projects
36:52are now being
36:53recognized
36:54and see very
36:55quickly that if
36:56you had a
36:57framework that
36:57was slightly
36:58more nimble
36:58and agile
36:59this is how
37:00you would do
37:00innovation very
37:02differently
37:02so it's happening
37:03and what I'm
37:04really pleased
37:05to see is that
37:06the EU is
37:07definitely leading
37:08on EU AI
37:09Act
37:09European Health
37:10Data Space
37:11trying to get
37:12consensus as to
37:13understanding
37:14what that trade
37:15off is
37:16if you give
37:16access to data
37:18what do you get
37:19in return
37:19how is that
37:20shared
37:20how is your
37:21health going
37:21to be improved
37:22so it's moving
37:23it's moving
37:24in the right
37:24direction
37:25and I'm very
37:25optimistic that
37:26it's going to
37:27continue in that
37:27way as long
37:28as it's clear
37:29and as long
37:30as it's understood
37:31and as long
37:31as people feel
37:32that they're part
37:33of this momentum
37:34and change as
37:35well
37:36thank you so
37:37much
37:37Alex
37:38you also work
37:39in dramatically
37:40different regulatory
37:41environments
37:42from China
37:43to South Asia
37:44especially
37:45some places
37:46that the
37:46regulation
37:47doesn't even
37:48exist
37:48it's non-existent
37:49how do you
37:50navigate
37:51deploying AI
37:52responsibility
37:53when the
37:55rules simply
37:56aren't there
37:56yeah
37:58it's a good
37:59question
38:00because I think
38:00even in
38:01very developed
38:03countries
38:03where frameworks
38:04are in place
38:05I think the way
38:06the technology
38:07is evolving
38:07for example
38:09right
38:09the number
38:09of updates
38:10that AI
38:11companies
38:11are pushing
38:12on models
38:12right
38:12it's actually
38:13more frequent
38:14than politicians
38:15meeting in parliament
38:16right
38:16so there's no way
38:18any existing
38:20framework
38:20will be able
38:21to fully
38:22quote unquote
38:23govern
38:23right
38:23the AI
38:24innovations
38:25out there
38:25but again
38:27I always go back
38:28to the first
38:28principle
38:29any tools
38:30whether it's
38:30actually AI
38:31or just new
38:31diagnostic tests
38:32that we're
38:33trying to do
38:33it needs to
38:34drive a real
38:36outcome change
38:37because if you're
38:38only intervening
38:40at one step
38:41in the process
38:42and you're
38:43not measuring
38:43the outcome
38:44right
38:44I think
38:45you're just
38:46fooling
38:46yourselves
38:47that you're
38:48doing something
38:48meaningful
38:49because if you
38:50truly believe
38:51that you're
38:51doing something
38:51meaningful
38:52then you
38:53should be
38:53measuring
38:53the outcome
38:54right
38:54I think
38:54that's the
38:56whole ground
38:56truth
38:56and so
38:57the way
38:58that we
38:58tackle it
38:58is
38:59what I
39:00just
39:00shared
39:01before
39:01right
39:02maybe
39:02a one step
39:02in the
39:03process
39:03right
39:03in the
39:04end
39:04we have
39:04to measure
39:05the end
39:05outcome
39:05I think
39:06the outcome
39:06is not
39:07just clinical
39:07but also
39:09economical
39:09and could
39:11have a
39:12very very
39:12hard look
39:13at health
39:13economics
39:15because I
39:15think
39:16healthcare
39:16in most
39:17countries
39:17actually runs
39:18under a
39:18budget
39:19it is not
39:20a business
39:21so you
39:22can't
39:23put the
39:24framework
39:24of traditional
39:25business
39:25building
39:26right
39:26when it
39:26comes to
39:27healthcare
39:27intervention
39:28because
39:28resources
39:29are always
39:29limited
39:30so I
39:31think
39:31if you
39:31trend
39:31and the
39:33way that
39:33we try
39:33to do
39:33we try
39:34to actually
39:34balance
39:34all
39:36not just
39:36the clinical
39:37benefit
39:37but also
39:38the health
39:38economics
39:39of it
39:39because if
39:39we get
39:40those two
39:40right
39:40we know
39:41that it
39:42by default
39:42will be
39:43scalable
39:43and if
39:44we actually
39:44do this
39:45right
39:45you're not
39:46actually going
39:46to be scared
39:47about whatever
39:48framework is
39:49out there
39:49that may be
39:50coming in
39:50the future
39:51because then
39:52we are
39:52confident
39:52that we
39:53actually meet
39:54not just
39:54whatever
39:55governance
39:55framework
39:56but we
39:56actually meet
39:58the actual
39:58real needs
39:59of the
39:59health
39:59systems
40:00thank you
40:01so much
40:01Anka
40:02coming to
40:02you
40:04you also
40:04discussed
40:05about this
40:06tension
40:07between
40:07rising
40:07technology
40:08ambition
40:08and
40:09declining
40:09system
40:10readiness
40:10in your
40:11article
40:13mentioning
40:14it's not
40:15just
40:15individuals
40:16that are
40:16left
40:16behind
40:17but it's
40:18an entire
40:18health
40:19system
40:19and
40:20that is
40:21due to
40:22unequal
40:23access to
40:23AI
40:23so in
40:24your
40:25point of
40:26view
40:26when
40:27regulations
40:28are lagging
40:29behind
40:29and
40:30AI
40:30infrastructure
40:31is
40:32unequal
40:32is
40:34an
40:34unprepared
40:35healthcare
40:36system
40:37by default
40:38will be
40:40unequal
40:41or
40:42is it
40:43going to
40:44steal
40:45function
40:46in some
40:46ways
40:49let me
40:50start by
40:50saying
40:51this
40:51I'm not
40:52going to
40:52say
40:52where
40:53but
40:54recently
40:55you can
40:56assume
40:56where
40:57that
40:57happened
40:58recently
40:59a doctor
41:00has been
41:01sued
41:02for
41:02following
41:03AI
41:04instructions
41:05in the
41:06same time
41:06another
41:07doctor
41:07has been
41:08sued
41:08for not
41:08following
41:09AI
41:09instructions
41:12this is
41:12not a
41:13scenario
41:14it's the
41:14reality
41:15in the
41:15system
41:16I will
41:16not
41:17mention
41:17however
41:19it's
41:19the reality
41:20we're
41:21currently
41:21living in
41:22so
41:23I would
41:24say that
41:24when
41:24there's no
41:25governance
41:26layer
41:27built
41:27into
41:28deployment
41:29accountability
41:30doesn't
41:31disappear
41:31it
41:32concentrates
41:33so
41:34it can
41:35concentrate
41:35on the
41:36clinician
41:36at the
41:36checkpoint
41:37it can
41:38concentrate
41:38on the
41:39under-resourced
41:41institution
41:41or at the
41:42patient
41:43outcome
41:43level
41:44and that's
41:45when it
41:45hits
41:46we have
41:47the EU
41:47AI
41:48Act
41:48in place
41:49well
41:50the EU
41:51AI
41:52Act
41:52requires
41:53since
41:542024
41:55human
41:57override
41:58capability
41:59what that
42:00means
42:00first of
42:01all
42:01that's
42:02law
42:02it's
42:02not
42:03aspiration
42:03but also
42:05it
42:06points out
42:07at
42:08the
42:09procurement
42:10level
42:10and that
42:11means
42:11actually
42:12that
42:12procurement
42:13it's
42:14still
42:14not
42:15enforced
42:16as a
42:16condition
42:17of
42:19purchase
42:19when we
42:21look at
42:22the parallel
42:22with the
42:23EU AI
42:23Act
42:24and how
42:24AI
42:24is purchased
42:25at
42:25institutional
42:26or at
42:27the
42:27system
42:27level
42:27so yes
42:29I would
42:29say
42:29an
42:30unprepared
42:30health
42:30system
42:31becomes
42:32inequitable
42:32by default
42:34because the
42:35risk doesn't
42:36vanish
42:36it does not
42:37just disappear
42:38it lands
42:39on whoever
42:40has the
42:41least power
42:42to catch
42:43it
42:43so my
42:44solution
42:45would be
42:45coherence
42:46and I
42:47would say
42:47that
42:48coherence
42:49is the
42:49intervention
42:50and
42:50everything
42:51else
42:52it's
42:52simply
42:53infrastructure
42:55thank you
42:55so much
42:56I was
42:57going to
42:57actually
42:58ask the
42:58last
42:58question
42:59about
42:59where
42:59do
43:00you
43:00think
43:00that
43:02like
43:02how does
43:03it
43:03basically
43:03failure
43:04looks like
43:04but
43:04due to
43:05time
43:05limit
43:06I
43:06actually
43:06want
43:06to
43:07turn
43:07it
43:07positive
43:07and
43:08I
43:08want
43:08to
43:08ask
43:09each
43:09of
43:09you
43:09to
43:10very
43:11briefly
43:1130
43:12seconds
43:13wrapping
43:14up
43:14about
43:15what
43:15do
43:15you
43:15think
43:16in
43:17one
43:17sentence
43:18would
43:18be
43:18the way
43:19to
43:19go
43:19about
43:20this
43:21issue
43:22of
43:23inequitable
43:24system
43:25that we
43:25have
43:25if you
43:26wanted
43:26to
43:26emphasize
43:26on one
43:27thing
43:27what
43:28would
43:28be
43:28that
43:29in
43:29any
43:30direction
43:30it
43:30can be
43:30governance
43:31it
43:31can be
43:31industry
43:32innovation
43:33research
43:33starting
43:34by you
43:35Joanne
43:35I'll
43:36be
43:36extremely
43:37brief
43:37and
43:37use
43:37one
43:38word
43:38which
43:38is
43:38transparency
43:42I'll
43:42always
43:43focus
43:43on
43:43the
43:44patient
43:44outcomes
43:46I
43:47would
43:47say
43:48very
43:48careful
43:49how
43:49we
43:50govern
43:50it
43:50and
43:51how
43:51we
43:51finance
43:52it
43:53include
43:54women
43:55in
43:55your
43:55team
43:55in
43:56the
43:56team
43:56that
43:56are
43:56building
43:57the
43:57model
43:57that
43:58you
43:58are
43:58creating
43:59right
43:59now
44:01thank
44:01you
44:02and
44:02I
44:03would
44:03say
44:03it's
44:04really
44:04about
44:04the
44:05choices
44:05that
44:05we
44:05make
44:06it's
44:06really
44:06not
44:06about
44:07the
44:07technology
44:07and
44:08if
44:09we
44:09have
44:10the
44:10incentive
44:11and
44:11the
44:12will
44:12we
44:13can
44:13always
44:13make
44:14it
44:14work
44:14thank
44:15you
44:15so
44:15much
44:16to
44:16all
44:17of
44:17you
44:17thank
44:17you
44:17so
44:17much
44:18for
44:18the
44:18audience
Comments