Discussion keeps the world turning.
For years, getting kids the right care meant heading to a big city hospital.
Local clinics perhaps just didn't feel trustworthy enough.
But now doctors and tech teams have joined forces to give community physicians AI backup systems.
You can think of those like a second opinion that never checks out.
Could this be the game changer that keeps families closer to home and still gets the diagnosis right?
This is Roundtable.
I'm Steve Hatherly.
Thanks for being with us today.
And for the show, I'm with Yushan and Xinyu.
First up.
For years, one of the biggest challenges in pediatric care wasn't the technology but the geography.
In many places, parents still travel to major hospitals because they worry that local clinics might miss something important.
But what if expert-level support could stay local instead of patients having to leave?
That question led to an unusual collaboration between doctors and AI developers, building an AI pediatrician designed not to replace doctors, but to help more children get quality care closer to home.
So could AI become the extra pair of eyes, if you will, that healthcare systems have been missing?
We're going to talk about that and more.
Xinyu and Yuxin, good afternoon to you both.
Let's start with the news, Xinyu.
There was a recent media report that came out talking about this AI pediatrician project.
Yes.
So this AI pediatrician project has been rolled out across multiple provinces in China.
So not just Beijing, we're seeing this project being carried out in China's like Hebei, Hunan, Henan, Heilongjiang, Shandong, Jiangsu, Shaanxi, Qinghai.
So a lot of provinces.
It's all over the place.
Yeah.
Yes, and so far it has been deployed in more than 30 medical institutions across the country.
And this project is basically a collaboration between Beijing Children's Hospital and an AI company.
And the goal was to tackle one of the biggest challenges in healthcare, which is the uneven distribution of pediatric resources.
Meaning that in urban regions, they tend to have more resources in terms of pediatricians, specialists, or medical facilities, whereas in rural areas, they may lack some of these resources.
That's a problem that's not specific to China.
That's something we see in many parts of the world where you have...
The majority of doctors or most of the doctors in the major urban centers, and then the further you get away from those urban centers, the fewer doctors and nurses and quality, perhaps, facilities there are.
The thing is that it is quite special for pediatric patients because children, it is quite hard for them to specifically or clearly claim what is the symptom.
And then parents can be even more worried when something happened.
And then if you cannot go to see the doctor really, really in time, you know, we can never know that what is the result.
Yeah, I think that's a good point you bring up because kids sometimes can't articulate exactly what they might be feeling or what might be wrong.
And then going to the local clinic...
I don't know if it's fair to say it or not.
I think so.
We had this in Canada, too.
My hometown is only 1,000 people, right?
So I remember when I was a kid, if it was super basic, we would go to the local clinic and see the...
Kind of town doctor he was the guy who basically took care of everybody but anything more than that and we
We went on the road the 50 almost one hour drive to the next big town to where the bigger hospitals were
But this isn't necessarily AI replacing doctors.
In fact, it's not that at all, correct?
This project aims at giving local hospitals access to expert-level pediatric support.
And the idea here and the goal here is that more children can receive quality care closer to their home so that they don't have to make that long trip like I did when I was a kid to a bigger hospital.
Yeah.
So in China, too, we have this kind of tiered health care that's recommended by the nation's health care department, which is that when your child has some kind of condition, you go to the nearest
Or community healthcare institution to take this diagnosis or treatment.
But if they cannot, if that condition is beyond their capability, then you're advised to go to district hospital.
And then if then they cannot also fix the problem, then you go to the bigger hospitals in other big cities.
But nowadays, because of these parents, they like skip the tier one and tier two.
They go straight to big hospitals like those in Beijing and Shanghai.
Yeah, because they feel like, well, I'm going to end up there anyway.
What's the point of going to step one and step two?
And that can make the hospitals in Beijing or Shanghai, they are overwhelmed by those non-local patients.
Well, that's just the mentality of parents.
I want to offer my child the best.
I can give.
That's why they're going to the best or the most professional doctors.
Yeah, and that's completely 100% understandable.
Any parent would want to take that step.
Yeah, so that can make the local hospitals, they are not having that many patients.
Even though they have the capability of fixing some of the common symptoms or conditions.
But with this AI pediatric project, like this example in Wei Nan Maternal and Child Health Hospital in Shaanxi Province,
It adopted an AI pediatric clinic in early 2026, so this year.
And they have this facility to assist human doctor in the outpatient department.
And it already had over 5,700 patients by the end of this May.
And the head of this hospital is hoping to expand this program to more than 90 primary and secondary level hospitals across Huinan City.
For those parents who are wanting to leave Hunan City to go to Beijing to have this long trip to have their children cured, they can easily have their children diagnosed in the city, in the local hospital.
The scale is quite big.
And the thing is that, of course, now it is the era of technology and artificial intelligence.
Um the hospitals they are also embracing um the way of ai can bring us
So i think of course we can see a lot of forms of hospitals using ai
Of course we're talking about ai doctors in hospitals but they are also like illustrating many forms
I think i saw one
That is they help doctors to diagnose or understand the symptoms of the patients
There are two screens one is showing the patients one is showing to the doctor and then and there is a speaker and a microphone and also scanner so you can scan the paperwork and the prescription
And also tell the symptom to the doctor and also the screen, and then the AI can help doctors to diagnose and understand better.
And other ones are probably like mobile phone apps, or they're just large language models that help a whole group of doctors to have a seminar on some very specific project.
This project is, well, it's kind of a mix of the new and the old.
And the reason I say the new is because the first generation AI pediatrician was launched in March of 2025.
So it's been just over...
A year.
A year.
And for many places around the country, they were adopting like the example you gave, Xinyu, as early as or as late as early 2026 and some still adopting now, I'm sure.
So that's the new.
But the old is that these AI pediatricians have been trained on decades of clinical records from Beijing Children's Hospital, and they've been refined with input from around 300 pediatric experts.
So that's what I mean when I say a mix of the new.
The tech is new.
But the cases were, quote unquote, old.
And that's what it was trained on.
Yes.
So this AI pediatric project is kind of aiming to build this comprehensive diagnostic knowledge system.
When it comes to the pediatricians, when they diagnose a child's condition, to some extent, they are diagnosing based on their experience or some of the symptoms or some of the test results.
But with this AI...
They are able to, you know, refer to some of the old cases in the past and how what are some of the common symptoms happening to this certain condition and they can make more precise like diagnosis.
They're also doing different language versions of it as well.
English, Russian, French editions are currently available and there are plans to introduce the tech to countries that are participating in the Belt and Road Initiative as well.
So it's going to go global too.
Let's get to the why now, I guess.
And we kind of touched on it briefly, but let's get more detail about it.
What's the reason that they see this as a necessary move?
Is it because...
And is it only because they want the parents to feel more comfortable about the quality of care they can get locally?
Of course, I think the resources and the shortage on pediatrician in China is one of the biggest reasons that a lot of these hospitals want to use AI to improve the efficiency.
Of all of the work and workload.
And some statistics from China's National Health Commission is showing that there are 162 children's hospitals nationwide and nearly 4,400 public general hospitals with pediatric departments.
And the number of pediatricians in China has grown significantly from 110,000 in 2015 to around 205,000 in 2024.
That's a big increase.
That is a big increase, around 74%.
But the things that the number of practicing pediatricians per 1,000 children rose from only 0.49 in 2015 to 1.02 in 2023.
Also very good, yes?
Even with it is, you know, improving, but there is still a substantial shortage of pediatric health care professionals across the country.
And I think I see the problem there.
So the national average number of pediatricians looks good on paper, but the doctors are not distributed evenly across the country.
So they would be heavily concentrated in perhaps a few big cities.
So the national average of 1.02 per 1,000 children, gosh, I'm trying to think of an analogy.
It's like saying that the average, if you think of the average temperature of a room.
Well, you could say the overall average temperature of the room is quite nice, but one of the corners of the room is on fire and the other corner of the room is freezing.
But together, when you look at the room as a whole, the temperature looks quite nice.
The average temperature is fine.
Yeah, exactly.
I don't know why I came up with this analogy.
The fire would be all the doctors concentrated in the big cities, and the freezing part would be the rural communities where doctors aren't evenly distributed.
That's probably a terrible example, but I hope we kind of understand the illustration there.
And the president of Beijing Children's Hospital, Ni Xin, he observed for many years that 70% of the patients coming to this hospital were from outside Beijing.
So he said that when he first took the role, he noticed that many parents, they are coming all the way from other cities or other villages to Beijing.
They were sleeping in hospital hallways or nearby basements.
Even public parks because they cannot afford the nearby hotels and they want to save money.
Because they've already made the big trip.
Yes, make the big trips and they want to save some money.
Sometimes it's even hard to get a room around the famous hospitals.
It's very crowded and a lot of people, they are lining up.
We can see that these patients and these worried parents they are making their decision to coming all the way to big city for better treatment
But now you know with this this kind of ai uh pediatrician project we are
Able to see that such problems can be solved and they can help local hospitals to build better capacity to treat some of the common conditions.
All right.
Well, then how let's get to the actual how this works.
How do AI pediatricians help in practice?
Because it's hard to kind of visualize.
We know what it's like to visit a doctor's office.
But I don't think we know what it's like to visit a doctor's office when AI is a part of that visit.
So what does this actually look like?
You mentioned it briefly about one screen facing a patient, one screen facing a doctor.
But what else is involved here?
I saw some examples online and doctors involved in the project in Chongqing explained that the AI doctor serves as a bridge between patients and healthcare providers.
For parents of pediatric patients, it offers, for example, first of all, 24-7 free consultations because they offer patients a kind of an app.
On the mobile phone.
So you can, first of all, before probably before you go to see a doctor, you can first tell the AI doctor the symptoms and they can tell you how serious the problem is.
And then you can decide whether you should go to see a doctor or maybe you should wait for maybe a little longer.
I mean, yeah.
I get it, but it feels- What's your concern?
That it's wrong.
You mean a diagnosis?
That's why I think the doctors are saying it is kind of a bridge.
So it is... Of course, you cannot go to see a doctor 24-7 or the specific doctor.
Yeah, no, I know.
I know.
And it's a good thing.
I just immediately thought... We know in the West of WebMD... WebMD is whatever...
The advice for people...
Is if you think something's wrong with you, do not ask the internet.
That is the worst place you can go.
Because the internet will tell you it's every possible terrible thing in the world is wrong with you.
And that's just what popped into my head.
I have a quite different opinion here because when you mentioned internet, it has the information from all kinds of sources.
Some are right, some are wrong.
But for the information that's being trained to this AI project, these AI tools, they have this information based on these things.
300 experts and their clinical records and all of the verified information.
So you can, I think you can trust the diagnosis given by more reliable
More than Google.
Yes.
And also because they are not replacing the doctors.
And that means the doctor will be the person ultimately give you the result or diagnosis, not just 100% based on AI.
Yeah, and it is a collaboration, and that's a great point, Jingyu.
There was an example from a women and children's hospital here in Beijing that uses AI during the consultation and also afterward to review whether...
The clinical reasoning made sense.
So the example was a child came in with a sandpaper-like rash.
They thought in the beginning that scarlet fever was going to be the diagnosis.
But then...
The AI took the test results into the system and then the AI suggested another possibility, which was mycoplasma infection.
I don't know what that is, but it's something that can produce a similar rash.
So then because of the AI's analysis, they did some additional testing and it turned out that that diagnosis from the AI was actually the correct one.
So for the doctor in the room, the AI didn't replace the judgment.
It just offered testing.
Another line of thinking which is why in the beginning of this uh chat i said another set of eyes yeah so to speak
Um which is always a good thing
You know doctors are incredibly talented capable uh intelligent people but it's never a bad thing to have another opinion
That actually reminded me of...
Qualified opinion.
Normally, they will give me medicine or I'll do some tests.
But the thing is that the whole process, it probably can last for months because the doctors are trying to give me different medicines to exclude the possibility of some specific diagnosis.
Really?
Sounds like an experiment.
It's kind of like that, yeah.
Because it's too hard to figure out.
Yes, there are so many reasons for coughing.
A number of possibilities, yeah.
And now with the, you know, because AI, they can search different answers and give you different possibilities.
And I think they are a good reminder and helper for doctors to probably ask different questions.
So that they can make a probably more accurate diagnosis.
It might be a mycoplasma infection.
And another example that I particularly like is this two-year-old child who had been diagnosed with cerebral palsy because the child couldn't walk.
And then the doctors using this AI pediatrician project, they gave this a different possibility that developmental delay linked into a sympathetic nerve issue around carotid artery, which means in the neck area.
So that eliminated the possibility of cerebral palsy.
And specialists from Beijing Children's Hospital later confirmed that the child actually did not have that kind of disease.
And after minimally invasive procedure, the child was able to walk continuously within a month.
Are we worried about the doctors here at all?
Well, in both of the examples we just gave.
The doctors were wrong with their diagnosis, right?
Or they were not that clear about the diagnosis.
Yeah.
So the AI kind of stepped in.
I know it's only two examples, but the AI stepped in and steered the doctor in the right direction.
I'm just wondering if the doctors...
Might feel resistant to this in some way because of ego, right?
Because it kind of undermines them
Challenge them
Do you see where I'm going with it
I'm speculating I'm just trying to think from different angles
But you know if you're sitting in front of a group of people or a family and you say as a doctor oh I think it might be this and then the AI says it's not that it's this
And the doctor is sitting there like, oh, right.
So this takes...
I think we have to give it to the doctors here that this is brand new for them.
This is something that they haven't learned to integrate into their own style of care, their own diagnosis before.
So it's going to take a lot of patience and a lot of collaboration, I guess, to make sure it's going as well as it possibly can.
And I think it is totally understandable that for both the human doctors and parents to, you know, not that 100% trust to these AI doctors, even, not even, like for me, I normally embrace technologies.
But for these kind of things, I still before the two examples, I really I was really cynical and I kind of still trust the human doctor.
For a lot of reasons.
But the thing is that in these cases, human doctors and parents, they also have mixed but also evolving attitude toward AI pediatric.
Yeah, let's talk about that then.
So what's some of the feedback that both the medical community is saying and also the parents who are bringing the children to these doctors in the first place?
Yeah, like Steve mentioned, some senior pediatricians, they are not particularly enthusiastic about this AI system.
Right, that's what I was, okay.
Yeah, part of it comes from their confidence in their own clinical judgment after years of practice.
And another part comes from practical concerns.
How is this AI technology really reliable?
Or sometimes it's too slow.
Because for the first generation AI tools, that's a screen.
And when a patient stands in front of the screen, the screen will ask some of the questions to the patient directly.
And then the doctor will just sit beside and listen to all of these feedbacks.
And AI will give this general diagnosis based on your answers that you've given.
But somehow, you know, to not that serious conditions, AI still ask like 15 questions or 10 minute questions.
And maybe for a human doctor, it can solve the problem like within three minutes.
And then that for them can take a longer time.
Yeah, exactly.
So this could be time consuming, more time consuming.
And that means fewer patients seen throughout the day.
If you're seeing, you know, 60 patients on a typical day, but because you have to sit there and wait for the A.I.
To ask questions and receive the answers and get the report ready, then that might cut that number in half and that's not good.
And also for that scenario, the doctor will think that I don't need that kind of AI.
I can solve the problem all by myself.
Or what am I doing sitting here listening to your conversation?
It's kind of what I was trying to get at before, yes.
Okay, so that's from the medical side of things.
What about parents?
Parents would feel at least a little bit speculative, I guess, because, again, this is pretty new.
Parents also have shown some kind of mixed reactions.
Some families are curious.
And willing to try the technology, others feel uncomfortable with the process because their concern is quite straightforward.
They're thinking that I come to see a doctor, not to chat with a machine, because machines are probably everywhere.
Many parents, especially in pediatrics, speed, reassurance, and human interaction matter just as much as medical accuracy.
So of course, the accuracy of these machines can be as good as human doctors,
But that kind of talk, that conversation with a real doctor is also very important.
And as we as we say, reassurance is something that is, I think, really important for a lot of these parents because they just want to have a very solid answer to what is happening right now.
Yeah, so that's the first generation of AI tools that we've talked about.
And now they've made upgrades.
So nowadays, the AI screen, they are not taking the dominance.
They are sitting in the background.
They're listening to the conversation between the doctor and the patient.
And they are not interrupting.
It feels weirder somehow.
There's a monitor.
Okay, so AI is quietly listening to all of the conversations.
Let's paint that picture though, because when you say it like that, it makes it sound like there's a humanoid robot in the corner, like staring from behind a curtain.
Not necessarily staring.
And also not necessarily a humanoid robot either, right?
Just screens.
They basically look like computer screens.
I think what they're trying to do is to clear the position of the AI and the doctor.
Because previously, they are trying to replace...
Some of the jobs of doctors, but now they're, I think, being positioned as a more of assistant or maybe after the appointment, they can have a better understanding.
Or as I said, a lot of doctors can have a whole group of seminar for these kind of things.
But that does address that concern, right?
You said parents were concerned about, I don't want to speak to a machine.
I brought my child here to speak with a doctor.
But when you remove, not remove, when you move the AI to the back of the room, as opposed to shoving a screen in their face when they sit down,
It's going to make it a little bit easier, a little bit more palatable, if you will.
But I mean, this can't be easy, though, because it is so new.
There has to be a number of difficulties here.
And I think we covered a few of them.
But I think another one would be we're talking about rural communities here.
And I don't know, but there could be a bit of a digital divide.
We talk about AI every single day on this show, so we have a pretty clear concept of how this might function.
But for a family in a rural community that doesn't really deal with AI that often, then it might be kind of a difficult concept to try to explain to a parent.
My question as a parent is, what is this?
And if the doctor says, oh, this is an AI that has been fed with 300 cases from Beijing, blah, blah, blah, blah, what do you mean?
What, right?
My point is, getting the parents on board
With this, with the trusting of this would have to be a big challenge, I would think.
And, you know, the thing is that pediatric visits are like, you know, rarely one on one.
And it is quite hard to really identify for these AI machines to identify who, which one is.
Is actually the symptom or which one is the children is speaking or the parents are speaking.
And for some of these areas in rural area.
Dialects.
Yeah, dialects is something it's also really hard for these AI machines to understand.
So as you said, probably that is, you know, like the AI or technology divide is something that we need to concern, but it can expand to many other areas.
That was the reason McDonald's got rid of their AI drive-thru thing, right?
Is because different accents.
Really?
They couldn't understand different accents.
Okay.
So they just decided to scrap it.
I think it was McDonald's.
It might have been a different company.
Yeah.
But instead of scrapping it, it should upgrade it.
It wasn't ready yet, I guess.
Luckily, for our country's AI pediatrician project, they are upgrading the dialect recognition capabilities so that they are able to understand more local dialects from these patients.
And also I think that another obstacle is funding because all of these AI devices, they can cost tens of thousands of yuan.
And not to mention additional expenses on maintenance and like upgrades to the system.
So these all can be for rural areas, they can be difficult to afford.
You think this is going to work?
The ultimate goal is to get parents on board to make them feel more comfortable so that they can trust the quality of care that they can receive locally is top quality.
As good as the care that they can get by going to a big city.
Do you think this is going to work or not?
I think that will definitely be helpful to both the doctors and patients because, you know, in Beijing's like big hospitals, they are already, they have been for years sending these specialists to the rural areas or to other less developed cities.
Yeah, but then the fear is that they're going to be there for a limited period of time and they're going to leave.
Yeah, but AI will stay there.
AI have the knowledge of these specialists and they will not determine your condition, but to help you.
Diagnose what your problem is and make suggestions and the ultimate decision is made by the human doctor that I feel like more assured
Hmm.
I don't know.
Probably still needs time because, of course, the thing the purpose is to solve some of the issue of the resources right for rural areas.
But while I need a doctor and now you are giving me a replacement.
So I don't know, is it better or not?
Because they did spend a lot of money.
They need a lot of money for investment and everything.
They're expensive, but are they really as good as what I expected?
I don't know.
I agree with you.
I think I think it is going to work.
I think it's going to work, but it's going to take time.
This is brand new.
And anytime we see something brand new, it takes a while to establish trust.
Do I trust an AI robot to make a coffee for me?
Yep.
Do I trust an AI robot to give me a proper diagnosis for what's going on inside my body?
I'd like to see some more people do that successfully before I'm on board.
But I do feel confident that we're going to see more and more stories coming out in Chinese media where
They're doing a great job, these collaborations.
You've got older doctors now who, like you said, Xinyu, are probably feeling like, oh, what is this?
I don't need this in my office.
But for the younger generation, the next generation of doctors, they're going to be completely comfortable with this.
And the next generation of patients is going to be completely comfortable with this.
So I think this is just the beginning stage, which is why we're having this type of conversation.
But I think it's going to be successful in the future.
