Yeah, because I think in English. If you speak to me in Mandarin, then it’s like two…

Double translation.

Double translation, so it’s a lossy cable. [laughs] Let’s just talk in English.

No problem. Basically, seven years ago, we started off in the market by hacking as many glucose meters in the market as possible, making data available to patients’ over a mobile App. With data, patients can then digest and better self-manage their conditions. As doctors have access to data, they can provide better interventions.

That’s awesome.

In the last seven years, we basically started to roll to the clinics in Taiwan. We project in the next two to three years we will have about 400 社區 clinics.

400 community clinics?

Right. Last year or two years ago, we had the pleasure of meeting Premier Lai.

Ah, William.

Yep. He organized a bunch of start-ups, and then Minister Chen Mei-Ling, we worked with NDC as well as the NHI. We did a project where we co-designed what we call a software development kit for any third-party apps…

That’s awesome.

…that would enable individual patients and users to download their personal data from the NHI agency.

That’s right.

The idea was to help promote My Health Bank by making the data useful and available. That’s one project we did.

You’re the My Health Bank?

We are not My Health Bank. We would be one of the third parties that leverages the data.

You’re one of the early SDK users?

Yes.

And you informed their SDK design as well? You’re like their pilot case?

We were. If you go to the meeting minutes, one of my co-founder who’s not here today, he co-designed the SDK with NHI. NHI outsourced the development of SDK itself. Yeah, we were the early-stage beta testers.

Awesome.

That’s one thing. What we are currently working on right now, and the opportunity that we see for Taiwan, is a new initiative called Digital Therapeutics.

Sure.

There are two reasons why Taiwan has this opportunity. Number one, Taiwan has good health care practices.

Single-payer.

And single-payer. The fact that it’s single-payer, data is available…

That’s right.

…even though data is in different silos. Maybe that’s on-prem. Maybe it’s in a cloud. Maybe it’s different pockets.

But it’s the same payer.

Yeah. What we’re doing now is, as we work with these healthcare providers, we work with patients, we’re able to capture, integrate this data as part of providing our service, and starting to build algorithms where that enables us to work with pharmaceuticals.

Why is digital therapeutics, why is data, why is digital service important in chronic disease management? When it comes to chronic disease management, it’s about managing behavior.

That’s right.

As an analogy, without data, there is no GPS to enable patients to manage their chronic disease. At the same time, for the healthcare providers, with data, they can have more precision in therapy.

That’s right.

I’m not talking about genetics. I’m not talking about DNA. I’m strictly just talking about by capturing…

Precision health, not precision medicine.

Right, exactly. 精準健康. Because of data, that makes therapy more precise.

More precise, more effective, but it’s not about designing a drug based on their DNA. It’s not that sort of thing.

Exactly. Even though potentially one day, that will come about, but that’s not what we’re doing. Honestly, what we do is not rocket science. Compared to what you do, we’re probably not doing rocket science. It’s basically capturing data, making data available, and then serving patients and healthcare providers with that data.

That’s the high-level introduction of what we do. I thought we could come and hear what you have in mind as our digital minister for health.

Sure.

By the way, you’ve been in the news a lot in the last two weeks. I’m here because…

I guess people from other jurisdictions wanted to make me an example to inspire their ministers.

Sure. [laughs]

That’s the main impetus.

Good, good, good, good, but basically, I would like to, first of all, meet you and understand what you have in mind as our digital minister when it comes to health data. How we can help, how we can play a role, and maybe how we can leverage your function to help us grow this global segment.

That’s the 30,000-foot introduction.

Which other jurisdictions are you operating in, or you’re mainly Taiwan-based?

Taiwan and Japan.

Taiwan and Japan.

Taiwan and Japan. Right now, Taiwan is our largest market. Our spot won 280,000 users in Taiwan, 410,000 users globally.

That’s right.

We’re predicting in about 18 months, our users in Japan will surpass that of Taiwan. Our largest investor is actually a Japanese insurance company.

I see.

They’re helping us grow in Japan.

There’s Japan and Taiwan, and then after we do well in these markets, then we’re going to think about Southeast Asia. For now, Japan and Taiwan keeps us busy enough, so just these two jurisdictions to focus on as a start.

In Japan, they have a legislation that enable the kind of data cooperative. I think they call it Infobank, or something like that, where people can pool their data together, join in the increased value that is only possible if you pool your data together, but have visibility and accountability into its governance.

If it pays dividends, then everybody also earns the dividends out of the act of pulling the data together. I don’t know whether you have any experience working with that new jurisdictional regulation. It is a new thing.

I’ve heard of it, but we don’t see it in action yet. In fact, in terms of data flow and the access to health data, not much going on in Japan.

I see.

In fact, a lot of our Japanese counterparts or partners in the…what we have in Taiwan.

In Health Bank.

In the Health Bank, and how third parties can access the data to create services for patients.

Ideally, My Health Bank is not just a SDK. It could be cooperatives. In Taiwan’s indigenous nations, they don’t start companies. They start a workers’ cooperative, where people share the fruit of their labor working together because it’s part of the indigenous culture.

True.

Then everybody has a say. It’s democratic. It’s not governed by who has the most shares, but rather it’s by community consensus. They elect their cooperative leader and things like that.

True.

I’m using that as a example, because before the invention of cooperatives, people thought that there’s a kind of unbridgeable divide between the capitalists and the employees.

True.

Cooperatives serve as the alternative example that prompt the capitalists to make more welfare adjustments and promote the labor unions to have a BATNA. [laughs] If they get fired en masse, they can at least form a cooperative, as a cooperative serves as a novel idea that bridged the previously undividable gap between the capitalists and the workers.

I’m currently seeing the private data thing, that there’s two different ideas going around. One is that the medical data belongs strictly to the individual, and the individual should have complete control, of course, also over how it’s used and things like that. If you have to collect all the consent, even for statistical purposes, it doesn’t quite scale that well.

Yes, scalability is difficult.

It hurts the overall community benefit. On the other hand, of course, there are jurisdictions where you don’t need to ask for consent. People just get volun-told to donate their data [laughs] to the state. There are other jurisdictions like that. Of course, it’s kind of effective to collect that data.

Then because people didn’t volunteer to offer, so data quality suffers and, of course, legitimacy suffers. Then, of course, there’s bound to be a lot of different accounts. People know that they’re being surveilled against their will, so they will maybe wear a mask, not for that purpose, for anti-coronavirus purpose, [laughs] but for…

Masking identity.

For masking identity purpose. It create an antagonistic relationship. Between these two ends of distrust, there’s bound to be some effort that can enable people who participate mutually in the data collaborative. In Taiwan, for environmental data, there’s plenty of very good examples.

There’s the AirBox, there’s the Water Box, and so on, where people volunteer and even donate to the distributive ledger of the PM2.5 level in their balcony or in their school. That enabled the Civil IoT project, which was one of the first projects that I proposed when I joined the cabinet.

For things unrelated to privacy, we already have a very good distributed ledger-based data collaborative arrangement. The thing is that this has not been applied to anything that is related to privacy, and…Go ahead.

It’s especially sensitive…

That’s right.

In Japan, from what I hear, the Infobank regulation is designed to create this third alternative, which is a great vision. Again, I haven’t seen any implementation. Because you operate in Japan, so I thought I would consult your opinion.

Taiwan is much more advanced in terms of activity when it comes to health data.

But we don’t have a regulation to enable it. It’s just the NHI.

It’s just the NHI, and NHI gets to decide, approve who gets to use the SDK.

Then legitimacy is lower, because in Japan they, like us, used to have data protection authorities distributed throughout the ministries. Because of GDPR, they are now also working on liaison office that can serve as a kind of final arbiter of such tanks.

Of course in Taiwan, that would warrant a independent unit in the administration. Then, later this year, there will also be one belonging to the Control Yuan, the National Human Rights Council.

Ideally, we should have a DPA in the administration, and a human rights oversee board that evaluates and approves NHI’s use to give it legitimacy. But at the moment, it is all administrative decision.

True.

Which necessarily mean a lower legitimacy, not to say initiated anything wrong. It’s just because it is not multistakeholder enough.

And not as efficient. The concept of the SDK, it’s individual, individual jurisdiction, individual legitimacy.

I understand that.

The scalability takes time.

Exactly.

It takes a service provider, like ourselves, to go out there and collect 200,000, 300,000 users, and so on.

I know. If any of the SDK users has a data breach, everybody suffers…

Potentially, yeah.

…because everybody will fought the SDK, not the user. Cambridge Analytica uses this way, and then FB plummets in legitimately. [laughs]

In your example, hypothetically, if we would be the one to be breached, we would be the one to be hacked, not the individual.

It has the same result. It has exactly the same result. Even by phishing or by scam, your user is convinced to give the same data they give you to some other black hat, even though your service is not at fault.

Once the black hat collects sufficient number of such individuals, who fall victim to the scam, they can publish that data and blame the SDK.

True.

Then everybody’s legitimacy suffers. Currently, we’re at a point where our technological advances is being hampered by the like of a legitimacy apparatus. If you ask what I’m thinking about, this is what I’m thinking about.

I see. Coming back to my position, in order to advance digital therapeutics, data needs to be made available to the service providers.

Certainly.

Legitimacy is extremely important.

Hugely important.

Consent in legitimacy. Integrity, legitimacy…

In constant counting levels, I can give a consent to be included in aggregate, or I can be consented to be included in a machine-learning algorithm, if I understand that algorithm and so on, but there is no consent of unlimited use, obviously.

If I cannot revoke my consent, I’ll be wary to give my consent, and so on.

True.

Currently, there is no legitimate way to label this kind of different uses.

For example, in vendor-rated learning, your consent is actually very weak. You just need to not actively be against your clinic sharing the wisdom, not the data, the wisdom, like the top two layers of neural network.

If you don’t opt out, you can be affirmed that none of your raw data will be shared, because it’s just the top two layers of a multi-layer neural network. This requires less degree of consent. The problem currently is that there is no equivalent of PM2.5 or AQI indicators for the kind of consent that we’re giving.

Of course, data need to be made available to you, but you can also give algorithm to clinic. Clinic runs one part and meet you midway. Currently, there’s a lot of different models going on. The problem is that the SDK, because it’s only administrative decision, there’s no clear, like organic food marker. [laughs]

Nowadays, organic food is what we call participatory certification. Everybody can go to the producer site and see that they’re actually using, responsibly, the farming…

Techniques.

…equipments and techniques. Currently, that mechanism is also lacking in the SDK scenario.

True.

These two points, one about the legitimacy before we actually run it, about the kind of consent we need and consensus to be reached. The second, while it’s running, a participatory auditing governance structure. These two need to be established before we can say that we need to expand this SDK to more users, so they don’t have to collect one by one.

Makes sense. Do you think this is something that the private sector can help?….Obviously, it won’t happen with the private sector alone. It would happen with the administration creating the infrastructure.

Exactly as we talked about, if people point to Taiwan’s work, it’s not about me as individual. It’s because they want to motivate their cabinet to do something Taiwan did. I’m just kind of a mascot. [laughs]

Poster boy.

Exactly! The thing is that I would very much like to say, “Oh, Japan has this Infobank. Look how well it’s done, and Taiwan should have the same.” The problem is I can’t really point to a specific example.

I see.

For this, a law change is needed, so we really need to convince people across all the four parties.

What can we do?

Find some examples overseas. People often point to Estonia. Actually, the mask sortition system that we just introduced has been running in Macau for a long time because they’re very small.

Before us?

Before us. Exactly, right.

Because they’re rich. They’ve very small in population, very small in area, and so they can afford to run experiments.

Actually, our NHI itself, the IC Card has been running the Pescadores in Penghu for extended amount of time before everybody in Taiwan gets the IC card, the NGIs card. For a while, everybody in Pescadores is using the IC Card. People in Taiwan is using…

A paper stamp.

Yeah, with paper stamps. We need to find our equivalent of Pescadores for this kind of arrangement.

Let me tell you what sort of assets we have.

Sure, of course.

As I was saying before, we have a few medical centers using our platform, but the majority of our healthcare provider users are clinics.

Right, so medical centers, like a dozen, or something? A few?

Five or six.

Five or six, OK.

This makes sense, because we’re not in the position of oncology, cancer, acute syndrome.

OK.

We are helping with chronic disease management. Naturally, this is business that the NHI is trying to push out, too.

Right. Of course, the NHI agency has all the incentive to go to the clinics.

Exactly. As a result, a majority of our software is deployed at the…

Clinics level.

Right. As a result of having a workflow, patient management platform at the clinic, we’re able to integrate pathology, self-monitor data, prescription data.

Right.

We’re starting to run algorithms.

Just to check my understand, the blood and urine tests, who keeps that data, in the machine itself or?

We create APIs. We do the heavy lifting with a different lapse, create APIs.

Right.

Build the pipes and then the water starts flowing.

I see, I see.

Somebody needs to go out and do this.

You’re like the de facto standard makers?

I wouldn’t say de facto yet, we’re trying to be.

That’s your strategy, anyway.

Right. As a business, that’s our strategy.

As the data starts coming up, we then run these algorithms and alert the doctors. Then the doctors make decisions.

I see.

Then when the patients use our app, they get education, they get support on managing diabetes. That’s what we do.

Now, before I go into the health data again, the whole idea about digital therapeutics is our number one chronic disease is chronic kidney related.

The what?

Chronic kidney.

Chronic kidney disease is the major disease.

Number one, and half of chronic kidney conditions are a result of diabetes.

I see.

Number three, spending, third, spending, diabetes. The fifth spending is hypertension, seven and eight, stroke and cardiac arrest. One, three, five, seven, eight, are all chronic disease related.

I see.

We’re trying to create a digital therapeutic that can make health care providers more efficient, save NHI money. That’s one thing.

Now, we were looking at a recent legislation in Germany, starting in April 2020, their equivalent of the FDA will basically approve apps and enable healthcare providers to prescribe the apps. This is the Digital Care Act.

I’m aware of that.

That’s what we’re trying to do in Taiwan. That’s one thing. As a result of capturing more data, then comes through my story of why we’re working with Sean. As we collect more data, these pharmaceuticals or these third parties, in order to do further research, would want access to that type of data.

Of course.

Then how do we give consent and even the economics as part of sharing economics back with the individual user…. That was the idea of working with Bitmark as a blockchain to capture legitimacy, integrity.

Kind of a grassroots legitimacy, investor chain.

Exactly. Those are the main two things that we’re working on.

One of the things that I’m working with the National Development Council and NHI on, maybe we can get your support as well, is this helping Taiwan create a digital therapeutics industry. Again, traditional pharmaceuticals, molecules, created in Switzerland or created in New Jersey, at headquarters, sold around the world.

Yeah, aware of that.

But because digital service is localized, is customized, that will not be the model like that of traditional molecules. There will be regional solutions, jurisdictional solutions that are created that can then be used in that region.

That’s right.

We think that Taiwan is a country that has an edge, because of what you mentioned 30 minutes ago, NHI, good practice, and the fact that we have data.

Yeah.

Those are the two things. The blockchain infrastructure on personal health data, and digital therapeutics is what we will continue to focus on. For however many years that may take us.

Right. This is very interesting. I’ve long been interested in this kind of data collaborative sort of ideas. But only with distributive ledger technology do it actually gain popular imagination. Of course, it is not everything, but it saves you trouble explaining things. [laughs]

True.

That’s very important, because people previously thought mostly in a client-server and very heavy-server and very thin-client mindset. If people keep thinking in that mindset, none of these consensus around data use would work.

True.

We need to have consensus before we collect consent. This is one thing.

The other thing is that I just recently learned about the My Health Bank team maintaining the health bank people. Because their system has been expropriated for a week. Nowadays, if you click…

Last night. [laughs]

Right. Starting last night, you can’t use any of the My Health Bank anymore.

Servers.

Because we expropriated all their servers to mask…

For masks, yeah.

Right, for a grand purpose.

Right, right. But it’s now working very smoothly. But they don’t have the capacity to provide their original services on the app anymore. Of course, it’s just for a week, and then we’ll delegate it to the app.

But what I’m trying to say is that, because I need to work with them, because I worked closely with them yesterday.

There really is, I would say, how do I put it politely, there is a lot of opportunity to create a more streamlined process around the backend processing of the My Health Bank experience. The current experience of going into the My Health Bank, although it’s explained very clearly, it resembles the tax filing experience before we did a redesign.

All the words are right. It’s legally correct, but that it’s…

Very hard.

Yeah, it doesn’t save us any effort to explain things. The second observation is that we need this kind of stuff. One of the reason why, yesterday, there was a lot of pressure, is that there were people using bots.

They just repurposed their ticket-buying bots, like for a concert. People would rush…

or train tickets.

…to be the first one. They try to issue 4,000 requests a second to try to be the first in line to guarantee that they have extra ticket to sell to other people. A lot of those bots were used against our system, which causes a firewall overload, not a application server overload.

It’s a denial service attack, I’m sure not by malicious people. It’s just people wanting to be the first in line, computationally.

Computation bragging rights.

Yeah, bragging rights, like, “I’m the serial number one.” It’s by sortition, so it doesn’t quite make sense, right? [laughs]

Yeah.

What I’m trying to say is that if we don’t explain the incentives very clearly, just like this poster.

If we say this in the very beginning, maybe people won’t try these bots against us that that much, because it really doesn’t increase their chance. Maybe we didn’t communicate this clear enough, or maybe they’ll just want to try their bots anyway. [laughs] That leads to the issue.

All I’m trying to say is that if the My Health Bank for each of the data flows has such a clear communication of the incentives of the mechanisms of what you can expect to gain…

Gain in return, yeah.

…in return, and the kind of control you can have. And if each of these one can be explained in this kind of…

Simple language.

…simple languages, I’m sure that it will save your time, and it will also save their time, because they don’t have to answer so many support calls anymore.

That’s what I learned from 48 hours from working with the Health Bank team. That’s the first time I actually work with the team.

Interesting. All right. Is there anything else that you think that this country should be focused on with regards to health data?

If you talked to NDC already, you know that GDPR adequacy is their top priority. Everything is about GDPR adequacy. Once we have a privacy act revised from the NDC…and the OPEN Data Act is a new act, they’re committed to push these two acts this year.

Through the legislators?

To the legislation. I don’t think anyone in the legislation will be against this kind of law to get GDPR adequacy. I don’t think any legislator would say, “Oh, we don’t want business from the EU.”

On the other hand, of course, legislators are free to add new clauses. I think that a public education campaign, once they have their privacy act draft, that would then have, as I said, a new data protection authority, a new DPA, that’s more independent as a organ, and also the enabling acts of what the GDPR called joint controllership of data, or what we’ve been talking about, data mutuals.

There’s many different names for that idea, joint controllership. If we can get these into the legislation language, and manage to make sure that all the four parties understand what this is about, I think we have a good foundation to start a public deliberation about what is the best model to go forward.

The current problem is that the same word mean very different things to different legislators, like the de-identification. This word in Mandarin means like 10 different things to 10 different legislators.

De-identification in Mandarin is what, 去識別化?

“Go and identify.”

Which is the opposite. It’s the opposite.

It sounds like a challenge, “go and reidentify this.”

ome people think putting some asterisks is de-identification. Some only consider statistics as deidentification, and there’s any number of things in-between.

If you ask a random legislator what de-identification means, if you ask 10, you’ll get 11 different answers. What I’m trying to say is that this is the opportunity for us not to use a piece of the language that is already so confused with meaning anyway.

This is a term where we need to translate the GDPR terms into Mandarin if we are to get GDPR adequacy. Then, just have a clear idea of what these terms mean, what exactly does a data controller mean, what does joint controllership mean, and things like that.

If we have the common language properly accepted by all the legislators, then we have a firm ground on which to the mechanism.

Then consent means something, [laughs] because otherwise I don’t know what I have consented to.

One last point. We took a look at the Presidential Hackathon.

Yeah?

We will participate on the health topics.

Oh, that’s awesome.

Serve individual patients better, based on the data that we can get off the user…

Right, identify right doctor for patients.

Or just write it into the treatment history.

Or the other way around. That’s right.

That’s something we will do.

Awesome. That’s SDG 3, right up your alley. [laughs]

Yeah.

That’s awesome.

Thank you.

You’re welcome. You can revise your Presidential Hackathon submission any time. Feel free to just submit your ideas.

OK. By the way, with regards to digital therapeutics, who do you think I should be working with? Obviously, this is NHI, but even before NHI, because, again, it’s digital. Is this something I can work on with your office or keep working with the NDC before working with NHI? What would you recommend? Again…

We don’t have a digital therapy command chain to the NHI at all.

Not at all?

Not at all. We don’t. Even in this work, the coronavirus work, we mostly worked with the CDC people on a working level. It’s just a technical level. The NHI has been expropriated [laughs] by us. We don’t really know their business logic.

What I’ve been focusing on, as I explained, is this top-level mechanism that need to be first ratified as regulation…

…and legislations before the NHI can run that. We’re at a stage what we call norm design.

…makes sense.

We don’t have a working-level relationship with the NHI, other than using their AP servers and firewall.

Got it. It makes sense. Once you have a mechanism to legitimize data, that’s when you can start?

That’s exactly right.

And then the imagination…

Can run wild.

…can run.

That’s exactly right.

All right.

We can keep in touch, and ST is in the loop in all this, including the NHI communications.

Absolutely.

Feel free to copy ST.

OK. All right.

All right. Thank you.

Thank you very much.

Thank you.

Thank you.

Can we get a picture with you?

Of course.