As the world continues to focus on slowing the spread of COVID-19 and determining how to safely restart the world economy, Expii Founder/Carnegie Mellon University Professor Po-Shen Loh has developed an innovative new contact tracing app that uses ultrasound technology to more accurately follow people’s exposure to COVID-19 than other apps that use Bluetooth alone. Join us to hear from Po and this exciting new app. https://www.cmu.edu/mcs/news-events/2020/0626_most-accurate-contact-tracing-app.html
Transcription:
Okay, good afternoon, everyone. This is Audrey Russo and I'm with the Pittsburgh Technology Council. I'm joined today with Jonathan Kersting. He's vice president of visibility and all things media. We are very excited about our guests today on a beautiful Pittsburgh day at the end of June. So I just want to tell everyone that there's a couple of things about the program. First of all, I want to give a huge shout out to Huntington bank. They have been with us from the beginning, and they work have worked feverishly as they always do to help small businesses and with all the loan options that have been provided at the federal level and state level. We have actually muted all your microphones. So just we make sure that there's no sound in the background. So to disrupt our speaker, and we have a chat. The chat will allow you to ask questions of our guests and we will Show to be interactive. Just don't use the chat as a way to sell your wares. That's all we ask today. So just a couple of things in terms of programming, we hope that you can join us tomorrow, we're going to have a very important discussion about the administration's recent suspension of several immigrant visa categories, and how that threatens America's position as a center of innovation and technology closer to home, how it threatens local jobs and opportunities for Pittsburghers. We're going to be joined by partner K&L Gates, who's going to talk about what the proclamation actually means. What can we expect in the coming months please don't miss that and share that with others. Okay, so in one moment, I'm actually going to bring on our guest who's potion low and he But first, we are going to show you the quadratic formula that has actually been put together which is algebra rap is a cool way to simplify if you can't factor get this trajectory showing on the graph the axis symmetry consider one of the best formulas to use in math on your work or to use a class that goes in like negative b plus or minus square.
I want to bring Po-Shen Loh on to the show right now. We got a little bit of a tastes to that. And I think he can find that on the on YouTube or anywhere. But first of all, thank you so much for taking the time with us. Thank you for being here.
Thank you real pleasure to be here.
We are actually going to eventually talk about COVID. But before we get there, I'd love for our audience to get to know a little bit about you, you have like a rock star background. And I'm very, very impressed with just you being here with us. So tell us about your love for math, your passion for social entrepreneurship, your programs like the hurt scholarship, and everything that led you to graduating first in the class at Caltech, to producing rap videos about the quadratic equation. So tell us, and you've got to set the floor now. So thanks so much for taking the time.
Well, thank you very much. I mean, I should say I just happen to be in the right place at the right time. That's how it always is right? And it happens to be that ever since I was gets kind of small. There were a few things I really liked him in life. One of them was I knew I wanted to help people. I like people. I like making people happy, like making people smile. I also like challenges. I like things that were supposed to be hard. And I also liked creatively coming up with some new solution, not necessarily just doing something, but coming up with something. And these were the three things I decided I wanted to do with my life. And it happened to be that math was a way that I could try to find those. And for a long time, I focused on math education. That's actually what brought me to Pittsburgh, Pennsylvania. It was Carnegie Mellon University, which was, I would say, a dream school, both for me as a student when I was applying to colleges, and also as a place to work because this was suddenly a hot hotbed of innovation and an innovative city. And when I came here, what caused me to do more of these things is I also thought that Pittsburgh was a really interesting place. As you heard from this intro, I like the challenges I like building things. And Pittsburgh struck me as a place where you could do this and in fact It was also an ecosystem where there might be many people collaborating. And so that's what caused me to start my social enterprise expii. And but expii has been doing for the last six years is to try to maybe change the way people can learn, whether it be through rap videos like you just saw, or whether it is through a free website that 300,000 people reach online, or whether it's through a line of online math classes, which we created for people who already felt the math that we're doing in school was too easy. So that was what I was doing. And I thought I would just be doing that online education stuff and stuff at CMU until COVID struck, and then suddenly, my life didn't go right and turn 90 degrees.
It did it really did. You were definitely someone who could see the future and you didn't you probably didn't even know that. So how did it change? Talk? Talk about that.
Right? So I will say I was actually one of the ignorant people in March. I thought that this was just a fluke. I actually was one of the people going around and saying I didn't know what it was in March. I just
I didn't know For me, right, right.
Sorry. No, no, keep going somewhere. Right? Right. So I was just I was I was one of the ignorant people in March who didn't know what was going on. And at that time, I saw that, you know, school started to close. And I was thinking, Oh, well, I guess we are providing online education. So we will just try to do what we can to help. And maybe what I will do is I will contribute through education. That's what I thought, except, except that what happened was that I also happened a long time ago to get my PhD funded by a group called the Hertz foundation. Hertz is the same Hertz as the rental car company, right? Unfortunately, many people are hearing about right now. But the founder of the founder of it did quite well and and created this foundation. And what the foundation does is every year they try to find 15 people across the entire United States of America to provide a free PhD in exchange for those people making a moral agreement to help if there's ever a moment of national emergency. And in the middle of March while I was preparing to run into the online education, direction, message came from a very senior and very respected person to this in this community, detailing how COVID-19 was not the flu. And that was a wake up call. For me. It was actually it was a scientific email explaining all of the situation that we now know to be the case. But I didn't know back then. And it concluded by saying, you know, if there was ever a moment of national emergency, it's not. And so that was this call to action. I did. I did, I did sign this moral agreement. And I happened to be in a place where I thought of an idea that maybe we could try to do, we could try to do this contact tracing. Actually, this was before contact tracing was even a commonly mentioned word. I'm a math researcher at Carnegie Mellon. My focus on math research is on network theory, which is the heart of contact tracing. Basically, from that point of view, I said, Oh, I see if we could just collect the whole network, maybe anonymously. If we could anonymously collect the whole network, we could help to prevent COVID-19 from wreaking havoc, and then that's when I said, Well, I signed this moral agreement I'm in and I just turned the rest of my life. To run into this, you know, you listen, you make it sound so simple and you make it sound so nicely, neat and tidy. But you had to have made some pivots in there. Did you feel nervous? Or did you feel like you knew what you were going to do? Or? I mean, oh, so I can tell you, when I started, I got this idea that, oh, let's just have the smartphones anonymously, see who is nearby and then send these alerts. And I thought to myself, this is easy. We'll be done in a week. That's that's exactly what you need. Right? You know, you need to, you need to think it's going to be so easy. So what I did is I got the idea. Maybe within three hours, I contacted the head engineer on XP, which is in our in our company, I said, hey, let's do this. It's gonna be so easy. And then I said, Oh, we might need maybe more than one other person. So then I posted out to my Facebook, which connects to a lot of Carnegie Mellon talent. And I said, Hey, you know, I got an idea. Let's go and do something that should be easy. And it was pretty soon it wasn't easy. But isn't that easy? Was it?
Oh, no, no, it's extremely difficult. Right?
What did you learn that you were surprised about?
Yeah, so so what we learned a lot? Well, first of all, one thing is once I decided I wanted to do something, I'm just gonna do it. So so you know, I thought it was easy, but I already said, I'm going to do it. And I'm doing it. Now, here's what we learned along the way. We learned that, for example, if you are using something called Bluetooth Bluetooth is how headphones connect to your your smartphone. I thought this should be easy to do. Every time I connect my headphones, they pair. It turns out that on the iPhone, that's relatively easy if you're programming the iPhone, but if you're programming Android, what I learned is that there are many different kinds of Android in the Samsung Android is a slightly different Android than the LG Android. And suddenly, if you also are supporting, for example, the Philippines we have people using us from the Philippines. They've got all these other brands of phones, which are all running some other Android. And so we found out that there are issues of trying to support a wide variety of devices. We also learned something else. We learned that there are there are fairly strict rules. What permissions Apple is willing to give in terms of what you can run in the background. So we actually, we actually cracked a lot of the technological nuts. By April 7, we actually cracked most of the technological nuts very early.
On one week, that was, that was that was three weeks. That was three weeks of not cracking their pedal. I love it. Well, I mean, the point is, the difficulty was how do you put together all of these technological pieces, we had something we actually were the first to release to the United States an app for download. And, and on April 7, but when I say an app for download, I mean, an anonymous contact tracing app for download. And when we when we released that we actually then had to overcome many non technical hurdles, regulatory hurdles, in order to make sure that we were able to release this to the to the world. So that would be one major obstacle. But the other major obstacle we found out was that even after you make something like this, it doesn't necessarily get picked up by the media. And that's not necessarily in our control either because the media can choose what to do, every journalist has the free right to choose what they would like to cover in their story. And so that's one reason why I'm very happy to be here on your channel. Because in some sense, we would love to be able to tell our story. And we wish that more of the world more of the US and more of the world knew about this innovation. That's, you know, it's coming right out of here, right out of Pittsburgh.
Well, you've gotten some exposure, right? Haven't you gotten some national notoriety or any kind of a press, we have gotten some Indeed, there was an article on TechCrunch, which went out. That's not bad. It actually detailed 15 contact tracing apps, and we are indeed one of those 15. What we are seeking to do though, is we are seeking to help people understand the difference, the difference between what we're doing and maybe what everyone else is doing, because we actually think that we might be able to help a lot right now, especially with this wave, the surging wave of Corona virus infections. That is happening all over the country. This is actually what makes me concerned.
So I met Stephen Colbert, once for a few minutes. So you want me to see if I can leverage that connection and get you on the show? I'd be happy to. Alright, that's my afternoon task, then I'll give it. I'll be excited to read it. Now. I like this. That would be great. So Alright, so listen, you go from being an accomplished mathematician, a professor, a social entrepreneur, a producer of rap videos to developing, you know, one of the top rated apps now for contact tracing. So you talked about the path. You talked a little bit about that. And then you look to develop this Nova app, and you see that as helping to change the world. Is that the right way to frame potion low?
I would say yes, I mean, the way I evaluate what I want to do with every day is actually how I can use whatever I might know to help the world in the most positive way. And so when I got the Do you know that I realized that this is actually the world's biggest problem right now? COVID. And that's because it's not only killing people, it's completely disrupting economies. And so that then that was going to kill even more people. So I see this as the central problem in our entire world right now. And therefore, the reason why I and the rest of our team are working so hard on Nov it is because we actually don't see any other potential systemic solution. We see a lot of people, you know, saying maybe let's stay home. But if you stay home, we also won't earn money. It's not gonna, it's not sustainable. Or you could hope that someone will make a vaccine soon. I think that we do need that it's absolutely necessary. But there's not really anything else that anyone has proposed as a solution. And so that's why as soon as we realized we could make this tool, actually, this would be a way that we could try to bring normalcy back. There's something I need to emphasize. We're actually not just a contact tracing. A lot of times people lump it in and say, Oh, it's like a contact tracing out like what Google and Apple are doing actually know, the the new feature, which is about to roll out is what we would call a COVID. radar. So what I mean is if you think about contact tracing, and I know there are a lot of business leaders who watch this call, contact tracing is basically a way of protecting, protecting the society by finding people who are already sick, and people who might already be sick and asking them to please isolate. Now, whenever anyone is contacted for that purpose, it's actually a not, it's not what they want. I would say basically, if you are pulled out by that, you are going to help the society and so you should comply, however, to you directly, it's not necessarily a benefit. And that's one of the difficulties with contact apps, right? If you tell someone, Hey, why don't you install this app, and the potential outcome is that it might cause your individual life to go a little bit down. Yes, you should do it. You should be a good citizen. But that's, that's hoping everyone's a good citizen. know if it does something a little different. Okay, what what we have, what we are what we actually voted to release, which leverages all the technology we've built, is something which lets every single person know how far away the infections are from them. Not in terms of geography, but in terms of physical relationships, physical interactions, here's why it's important. Suppose that you live in the in the South Hills. And suppose that you work in, you know, in the UPMC tower in downtown. Does it matter as much for you whether or not COVID-19 has gotten to dormant? Or does it matter whether it's gotten to your floor in the UPMC? Building, right, even though you live in the South Hills, it's actually not the geography that matters. It's who you interact with. And since the COVID app actually collects all of this anonymously, well, we don't know who is who that's the entire point. We don't know who is who. We just know that your phone has some ID. And the ID doesn't tell you anything about the phone. It doesn't tell us anything about the phone, but we know that ID number 15,000,002 106 was with ID number 20,000,305. And then as we see COVID spreading, if we see COVID, spreading to somebody who has maybe five degrees of separation from it, think like LinkedIn, think like LinkedIn, but in person, right, then we will tell you, it's already gotten to this distance of fact. And what this gives us a value proposition to every person is, you're not sick yet. You actually don't want to be sick, selfish self preservation. I actually personally want to install this app, not because it's going to make me have to do anything. It's so that I will know when I'm going to work remotely for a while. You see what I mean? What I'm trying to say is that the entire incentive flips because suddenly the paradigm shifts, it is not that we are trying to find people who are already sick or who might already be sick and asking them to isolate. No, we're trying to help every healthy person stay healthy.
You flip it on the other direction, you're inverting it and you're inverted making that's actually the that's the new development is this available now I can download this app, the NOVID App is already available and what it's doing, as soon as you download it, it's going to collect that network for you without telling anyone who you are. And we expect that next week, we will actually be releasing the version, which suddenly you'll see something new on your dashboard, which is How far away are these positive things? But that's, of course, based on the network you're building. So when you start building the network.
So we have we have a great question from Larry, who works at PwC. So Jonathan, do you just want to read it, or I can read it. You can read it, Audrey, go for it. Oh, this is what Larry joy is saying. He's saying can you speak to the thinking behind the how and why your team went? With the ultrasound technology versus other more well known wireless methods ble being the most common, but there are other technologies that leverage ambient signals in similar fashions for anonymous contract, tracing Wow,
I would love to take that question. And the reason is because I just stayed up all night doing something.
Okay, good. Okay, Larry of PwC.
Yes. I have a question. Am I allowed to share my screen? I tried to press share, but it said host disabled screen trying to share. I'm trying to share a PDF, which I just made by staying up all night. I'll find him. You continue to do that now.
Oh, it appeared it appeared. I'm good. I'm good here. Okay. Here it is. This will answer Larry's question. This should so let me see. I have a question. Do you see a graph right now? Yes. Yes. So I stayed up all night doing this. And I'm not I'm not kidding. So so so what happened was a minute late for this cup. We actually have just conducted an experiment with 600 measurements. And this is showing the level of accuracy of the ultrasonic method. If you see this graph, this, this this line right here, the diagonal line is the True Match. between an actual distance and obits measurement, oh, sorry, sorry, that line is the ideal. You wish everything was on that that diagonal line. Now the actual dots are measuring different settings of using Nova. The important thing to notice is that for actual interaction distances within six feet, six feet is 72 inches. If you're under the six feet, the cluster of data points is extraordinarily tight around that line, even up to about nine feet, which is 108 inches, it's also extremely tight. And, I mean, I don't know if you can see from the graph, the important thing is that there is this line, and then all of the actual detections are above the line. So let me just explain what this outlier means. This is a triangle way up at the top at a distance of an actual distance of 150. The top triangle says that, you know, in a particular setting, the two devices were 150 inches apart, which is about 12 to 13 feet, and the devices thought that they were approximately 300 inches apart, which is this one Math is actually too hard for me, it's probably like 28 feet or something. I probably did that wrong, but something like that. Is it 2824 24 feet, something like 24 feet 2425 feet. Now, the important thing about this graph is that almost all of the dots are above it. So as you know, if you are implementing any kind of a contact tracing solution, people are very afraid of false positives. People don't want to be told you might have been close to somebody, but you actually weren't. And if you if you tell someone that then you might be asking them to stay home for something that they actually aren't needing to stay home for. The value of ultrasound is that it actually always overestimates always overshoots so if we actually use this to categorize the other fun fact, on this piece, let me let me see if I can scroll. Yes. The other fun fact is here. This there's this graph here. True distance, number of samples number correctly categorized percentage. Actually, out of the settings we measured. We measured 402 settings. Where the distance between devices was greater than 12 feet, that would be a situation where you don't want to tell the people that they were close together because they weren't close together. And the issue with the other things you've mentioned, like the Wi Fi and the Bluetooth is that they can't tell if you were, you know, if you were 12 feet apart, they wouldn't be able to know where you're really 12 feet apart are less than six. Our data is that out of the 402 samples that were more than 12 feet apart, 401 of them were correctly categorized for an accuracy of 99.8%. That blew my mind. And now of course, there's a way to get to high accuracy that way to get a high accuracy, if you want to categorize the samples that are far apart as being far apart is to just say, all things are far apart. That would be an easy way to get 100% score on the greater equal 12. But the less than or equal to six line shows that you know, we didn't achieve 99.8% just by simply turning off our detection and telling everybody you are you are you're far apart. You're far apart on the ones that you were actually close together out of a hole. Hundred and 87 samples 103 were correctly categorized 55% might sound like it's very low. However, if you have no app, your percentage is zero percent. So but the point is, if you can capture half of them, that's actually a really a big deal. I mean, there are more details in this particular particular research study. But in this particular research, study paper, this paper you released, let me pull this back. I just did a stop share. We are actually attempting to release that paper right now.
When you get it ready to be released, can you share it with us?
I would, I would be honored to actually what we are doing right now. Again, it's because this is this is hot stuff. Hot meeting is just hot off of the off of the research. And so what you just saw is the write up I wrote last night, and I'm sorry, I'm glad you stayed up all night. Oh, it was exciting. I mean, when I saw 99.8% That's insane. It was worth it.
Because the question that was Nobody using any of those technologies that you just mentioned, would even dream of trying to find 99.8%.
So let's just flip this for a minute now. Now let's talk about privacy. Oh, yes. Okay. So I mean, we're concerned about it. We've opened up our world over the last decade. And now we're saying, Wait a second, what? What have we allowed? And where does this fall into that? I mean, you're working on social entrepreneurship, and how it's now my privacy is totally being blown up under this model.
Right. So this is one reason by actually at this point, we haven't made any contracts with companies yet. And this is where I think it's great that there are so many companies represented here. Because the reason we haven't made any contracts is because we didn't feel comfortable exposing to companies a linkage between your user ID and something in their database. Right now. The way that we run this is that we actually do don't know anything about anyone. And that's because when you install the app, you don't tell it your phone number, you don't tell it your email, don't tell it your name. It doesn't use your GPS, it actually doesn't even tie anything to an ID on your phone. It actually, it also doesn't record, it doesn't record your IP address either. What it does is it just instantly generates a totally random four digit, 40 digit ID number. And it doesn't even tell you the reason it doesn't tell you so that your employer can ask you to give it to them. So I mean, not not just the employer, but even the police. If the police stopped you on the street and said, Show me your ID apps user ID, the answer is it is impossible. And so what we do is you see, you can think of what we've made as like an anonymous email system, where as soon as you install the app on your phone, your phone is constantly checking back to phone, which is our servers saying, hey, my ID is 50 quadrillion 270, whatever. Do I have any new messages from the system? It doesn't say this is the system can't go and say I want to contact Audrey's foot We don't know. It has to be that Audrey's phone phone home and said, Hi, I'm using number this like enormous, crazy thing, which Audrey doesn't know, either. Do I have any new messages. And by the way, I just recently saw in my vicinity, there was somebody who was going by the user ID of like 40,000,207, please record that we were together. So that's what we're doing on privacy side, on our servers. We don't want any personal information. We don't want any health information. And that's why we've had a little bit of difficulty finding a solution that would work with businesses, because many businesses actually run it, they wanted us to say, Hey, you know, just tell us the linkage. But to tell us the ID, we want to link it to the ID number of the employee so that I can tell this john smith, john smith, you were next to so and so you need to go home. And since we couldn't match that we actually didn't release that. That's why we're a social enterprise.
Well, first of all, I hope that your ideas and your discoveries here on the data privacy side can be applied to all the rest of The apps that are actually many of us are addicted to. So I feel like I wish that for you. But I also want to know as we sort of wrap up is what's next for you? What's it? What are some of the challenges? What's, what's your roadmap, in terms of social entrepreneurship and this NOVID app?
Right? So I started in education, it looks like I ended up in healthcare, but I wouldn't call it ended up, I'm the kind of person who's always looking on what I can do to try to help as many people as possible. After getting into this contact tracing, I call it anonymous contact tracing, we actually see that there are going to be many applications of this beyond COVID, even in other countries in the world, where there might be other kinds of pandemics that we don't necessarily hear about in the United States, but are important to stop, for example, maybe tuberculosis. So I actually see that the things that we are doing are generally enough that they could actually contribute to human health. So I would anticipate that until COVID blows over, I anticipate that we will be deeply involved. If nothing else, for one reason And the reason is because why we are pushing so hard is we actually at this point still do not see any other legitimate solution that might stop the spread of the disease. Even all the other apps that we see, we don't think that they're as effective. And so since we don't see any other thing being done in the entire world that could attempt to rope in this COVID-19, we will keep fighting, fighting to stop the COVID-19 ai. And you know, we might end up being a group that starts coordinating some sort of pan globe response, just through the fact that we happen to operate the free app. And that's why it's a social enterprise. And we didn't start by saying, how do we monetize this by selling to companies? In fact, companies came to tell us, we'd love to pay you for this. And what we said is, you know, we're just gonna keep building the demonstrably best technology in the world. If at some point we can see eye to eye on the privacy concerns. We can do a contract with you. I'm not I'm not against that. But it has to be privacy, integrity on that and it's very easy as you build a company to lose your way and If you can help us, you can help the world not just with COVID and contact tracing, but actually with privacy and holding us up to those kinds of ethics and those kinds of standards. So if I want to to audit a class from you over at Carnegie Mellon, Hmm, what class are you teaching next? Ah, so the class I teach next fall is a class called How to solve problems that you have no idea how to start.
Oh, so I could probably audit that class.
Actually, that might be possible. So what I did in the last semester, because everything went online, it's a YouTube livestream everything. So last semester, my PhD level class was running on YouTube live every single day, and they were random people from around the world just watching. Actually, I mean, I don't know if you noticed, but every day from four o'clock to five o'clock pm eastern time, I livestream an interactive math session on YouTube, where random people from all over the world asked me math questions, and I make lessons on the spot.
We need that link. So give us that link. And then I can take the link to to Stephen co bear. Ah, if we can get him on that show.
I think that'd be great. Very cool.
Listen, you can listen more to Po because Jonathan Kersting did a really lovely deep dive with them. And we can share that link with you. And I think you'll really have a deeper appreciation for the man, the vision and for the work that he calls easy. Everyone said. Well, we are thrilled to have you. Thank you for taking the time. Thank you for loving Pittsburgh, thank you for loving Carnegie Mellon. Thank you for teaching and sharing. And thank you for caring about the issues around data privacy, because that is absolutely real. And it's so tempting in the time of COVID, for privacy to be actually, you know, overhauled, and my hat's off to you. So thank you. Thank you, everyone. I think there were a whole bunch of messages that I I think we missed an I honestly we've got a we've got a log off we'll try to capture some of those and see if we can get answers to the questions. Po's the National coach of the United States international math Olympiad team. And under his coaching the team won the competition in 2015 2016 2018 and 2019. And their first victory since 1990. Whoa, you get that right. quickie. action packed. Thank you, po we're gonna we're gonna stay close to you. Okay. And thank you for really great. Thanks, everyone. We will see you tomorrow, same time. Enjoy. This is Audrey Russo joined by Jonathan Kersting from the tech Council. Thanks for rocking it Po. Thank you.
Transcribed by https://otter.ai