Session is finished.

Jump ↓

.
[SPEAKER] I'm Emily Berry from the Miami University Alumni Association.
And it's my pleasure to welcome you to this session of winter College 2021 for more than 17 years, Winter College has been the alumni associations, premier, alumni education events we're so excited.
To bring it to a broader audience and our virtual format, we have an amazing lineup over these two days, you can navigate the full schedule by clicking events by type on this one, toolbar, and slipping winner College 2021 from the drop-down menu.
Feel free to join programs even while they're in progress.
If you can't make it's all of them sessions will be recorded.
Then posted online.
Our session this morning is foresight.
The science behind the government's efforts to predict the future and protect the people presented by DJ Rao.
Doctor [inaudible] Rao is an associate professor in the computer science and software engineering department of my Amy's college of engineering and computing he teaches operating systems and hyperplane, competing courses in the department and leads the Grand Challenge.
Scholars Program.
Dr.
Router seem to be in Computer Science and Engineering from the University of modern India.
And his MS and PhD in theatre Science and Engineering from the University of Cincinnati, let's get started.
Remember you can submit questions or comments during the lives.
By clicking the link below the video.
Good [SPEAKER] Morning, everyone.
Thank you so much for taking the time to be here with us on a Saturday morning.
I'm super excited to be here with all of you.
And share some of my experiences and some of the research that had been doing here at Miami along with my undergraduate and graduate students for the past few years so let's get started today before we get into the details, just a quick outline.
I would like this for this to be more of a conversation than anything else So.
Feel free to ask questions.
So anytime you have a question, put it into the chat, and then I do have blocks there, I would like to answer questions, but I would like this to be as interactive as possible that we can to try and simulate a class like environment there.
You can raise your hand and ask a question at any time.
So more questions, the better it is for me and the better I'm able to connect with all of you and have this to be more of a conversation rather than a presentation so I just want to introduce myself and then talk a little bit about my research and the motivation for the government works with the manage epidemics, which is one of the pertinent topics at this time and then we look at a high level overview of some of the research that I'm doing and then we'll wrap it up so please do anytime you have a question, please feel free to post these questions on the chat and I'll be happy to answer these questions.
Just a little bit about myself.
I graduated from University of Cincinnati with a doctoral degree in 2002.
I've been in this area for a long time.
I worked with a company called Elsevier that's housed in LexisNexis, little bit down south of Dayton, Ohio.
And then I've been working at Miami on and off.
I work close like 14 to 15 years now at Miami.
And most of my research deals with paddle and Distributed Computing Applications of supercomputing and such so I work in different areas like epidemiology, which is the study of diseases with bioinformatics this is to analyze different kinds of biases and now I'm slowly starting to get into design like urban planning into smart cities and such.
Here's a picture of my daughter and wife My daughter recently won a HOSA competition.
This is on the health and such as a picture taken when she got her a worn at the host a competition.
My son works for Google.
He specifically works on his protocol, YouTube, and his job is to put up cat videos on the internet for people to watch.
Just kidding, he does more than putting up cat videos.
But you guys know what I'm talking about on YouTube, here's our other child, Zeus.
He's a dog and he's always here with me while I'm working and keeping the company.
Necrosis, we also have a cat.
His name is mochi.
So this whole family unit keeps me busy along with my research and such.
And in addition to doing my research, I also serve as a director for Miami's Grand Challenge Scholars Program.
Where we're focusing on some of the most complex issues facing our nation.
And to be able to motivate and work with our undergraduate students in different areas of science and engineering to be able to address some of these challenges and this program is supported by the National Association of engineers.
So when students work on these programs, they not only build a multifaceted skill set in terms of research service learning, leadership, global dimension, entrepreneurship they also get recognition from the National Association of engineers, which is a very, very unique recognition that helps them showcase their skills, countries.
And sets them up for future success.
So this is a program that we had actively doing.
Now at Miami.
It is predominantly housed at the College of Engineering and Computer.
But all students in Miami University participate in this program.
If they would like to sew and this is a program.
We'd also like in woman from alumni and people who are working in the industry to give ideas to students so if you're interested in some of these social and other issues that are facing our nation.
Please feel free to reach out to me and we would love to have you come in, engage with the students and have a Congress.
Would them and spur new ideas and thoughts with the students.
All right, with that, let's get into the topic at hand, which is the population growth and urbanization that we are going through, is increasing the density of people on planet Earth.
And because of that, and due to all of the air travel and transportation, there are lots of challenges that we face, which is increasing risks of epidemics in populations so you get back in 20142015, particularly in the Midwest, we had a bird flu epidemic there, over 50 million birds.
To be called this is by calling the birds is the only way we can manage, currently manage these epidemics.
And that causes a lot of loss.
Over $4 billion in losses were these poultry farms and such.
Then we had a chickungunya epidemic didn't impact the United States that much.
But Central and South Americans were heavily impacted.
We had more than close to 15 million cases and that's still an ongoing issue in the Americas.
And then we had the Zika epidemic that came out over a million cases that were reported.
And now we have this COVID epidemic that's ongoing..
We have exceeded over a 100 million cases.
And over 2 million people have died.
So these epidemics had been growing and there has been the steady stream of epidemics that we have, been facing and this is going to be a challenge as we move forward in the future so just to give you an idea of these epidemics, the avian influenza particularly was in the Midwest, impacted almost a quarter of a million farms in the Midwest and almost 45 million turkeys.
And about 50 million chicken culled.
So you have to utilize these birds because the birds are housed and pretty dense.
Housing.
So there is an infection.
All they can do is literally the truck load of these birds.
They die.
Some of them just died due to the bird flu.
So all we can do is take these huge trucks, make a giant pit and bury the birds.
And that causes a lot of losses.
Economic losses where you lose these millions of birds.
In a day.
And that causes serious problems.
And of course, we have lot of mosquito borne diseases.
Here is Bill Gates, apparently when he goes to a presentation, he travels with this little jar of mosquitoes.
These are homeless mosquitoes apparently.
But still it's interesting to see.
Bill gates traveling with the jar of mosquitoes, Maybe sometimes I being I should be travelling with the jar of mosquitoes Just kidding so mosquitoes are one of the big challenges one of the big backers of diseases these days so this chickungunya epidemic was the initial one This is where I started engaging with these trying to work seriously on these epidemics.
And trying to be able to mitigate the impact of these epidemics.
And here with this chickungunya epidemic, it's just started on a handful like of 4 Caribbean Islands, a few cases is what it started with.
The mosquito-borne disease.
And then soon it exploded in the Americas.
And we soon had within about a year.
We had 1.5 million cases and the virus had spread to all 51 countries in the Americas and this was a kind of eye-opening epidemic for a lot of people in the sense that most of the population was naive, didn't have immunity against the chikungunya virus.
So when this virus came up, it rapidly spread and established itself.
So this was a somewhat of an alarming trend because we had handle on mosquito populations over malaria, dengue epidemics but it turned out that we did not.
And then that's followed up with immediately overlap with the Zika epidemic again, a mosquito disease.
And the challenge with Zika is almost 80% of the cases are asymptomatic.
That means they don't have any symptoms, but they're actively transmitting the infection to other people.
And one of the big issues with Zika is it causes microcephaly in babies.
So if the if a pregnant mother gets infected with the virus there is a chance that the baby will have problems with brain development costing Microsoft and that's a life long problem that you will have deal with so it's not like a short-term problem, but it's a long term socioeconomic problems that people face and that puts a lot of hardship on people.
And of course we have the ongoing Covid pandemic.
We're all now acutely familiar with this pandemic.
We have had over a 100 million cases worldwide.
And we have had over almost 2.5 million people who died from this pandemic.
And it's known the world and it's a global pandemic lot of countries have been impacted with this and of course, I have had a personal impact with this pandemic last year, my dad got a Covid infection and he passed away so this code pandemic really hits close to home for me all of these pandemics.
So clearly these pandemics are a big challenge and it's impacting a lot of people, not just in our country, but worldwide.
So the real question that a lot of the agencies, government agencies, public and non-profit organizations are dealing with this.
How can we plan better to mitigate these epidemics?
How do we define effective processes?
That means, how do we reach people in emergency?
And how do we mobilize resources how do we plan for vaccinations?
How can we build the capacity to manufacture these vaccines and distribute them?
And all of these are challenges that we need to collectively solve so that we can protect not only our health, but our way of life over socioeconomics not just in our nation, but worldwide because we now live in a global community, we are no longer isolated little countries.
We are globally interconnected.
So these kinds of big missions of big ideas need to happen at a global scale.
Questions so far.
Maybe you guys can share some of your experiences with the pandemic.
You guys are welcome to share some of the things that you may have found unique.
Living through this pandemic and we can have a conversation about.
You can put those questions in the chat box on your on your webinar and I'd be happy to answer those questions.
Any questions up to this point so we spoke about why all of the different pandemics that we face and why we need to address them.
We don't have any questions yet [SPEAKER] If you guys have questions, you can post them at any time on the chat.
And then I'll be happy to answer those questions.
Good.
So these epidemics have serious, Health Related socio-economic impacts, they are typically multinational affects multiple countries usually.
And they cause a lot of threat to national security and stability so over the years I worked with several agencies, were actively involved in trying to mitigate these so I worked with the Centers for Disease Control and Prevention.
I worked with the Defense Darpa, defense advanced research project agency.
I worked with the Department of Homeland Security and also the one of their subsidiaries or one of one of my clients was also the border.
Bureau as well.
And then I've also worked with the Office of Science and Technology that is housed in the White House that directly reports to the President of the United States as well.
So a lot of these agencies are collaborating in trying to mitigate these epidemics.
There was one question, Oh, I'm getting questions from people awesome.
Excited.
There was one question from Michael what is the realistic time frame for covert to subside?
That is still projected out nobody in the scientific community is still willing to wager a, even a time window yet.
But the word on the street, so to speak, of the instances, not a scientific consensus but the word on the street is it's going to be another season, like maybe only towards the end of this year, end of 2021, maybe we'll be able to have a more realistic feel for when this COVID pandemic is going to subside.
Thanks for that question Michael.
Maryanne.
Thank you for your question.
There was a lot of time plan for vaccination distribution why does it seem so chaotic?
No good question.
And we look at this challenges because vaccine, the, challenge with vaccines is safety that means it takes our time to manufacture a lot of these vaccines and then the production of these viruses, these viruses are hard to grow even though it may seem like they easily infect us.
Actually controlling the culture FOR these viruses can be very difficult in the growth medium for these can be difficult.
And that's why now we have some COVID vaccines that have to be kept at such a low temperature that even distribution at those low temperatures becomes hard.
And so vaccine manufacturing is not easy.
And then these low temperatures make a challenge in distribution, storage, administration.
And then we need these tests to understand both short-term and long-term.
Effects to make sure that these vaccines are safe.
So because of that, producing manufacturing, planning, distributing these vaccines is definitely a challenge and you need leadership for this let me be very honest about this you need leadership at every level in the system and then you also need It's not only, in fact, I would say sometimes it's not even an engineering problem it's more of a psychological philosophical problem.
We have to convince people to take these vaccines just because we have vaccines doesn't mean people are going to accept these vaccines and protect themselves.
So it's a multifaceted where we need this in our whole society needs to come together in a meaningful way to address these these are big challenges, right?
Thank you, Dave.
Thank you for your condolence, Dave.
It WAS a BIG HIT.
My DAD WAS 77 in pretty good health and then he was actually outside distributing food FOR some there was some food shortages he lives in India, so there was some food shortages.
He was helping out with distributing food and such and somewhere along the way, he was protecting his bedding glossy or wearing masks and all of that.
But yet he picked it up.
So that WAS that WAS a BIG head for my family.
How did you first get involved with working in the US government?
Yeah.
I got involved working on a darpa challenge for predicting this chikungunya epidemic.
And that was and that work came out to be really good.
And I won the best methodology award from darpa for it.
And that's when the invited me over to the White House and they were talking and I was also involved in some of the Ebola's R0, because that's where the columns are, don't ask me why.
But I was also involved with those.
I kind of kept working with the people and that collaboration has continued over the years.
What has biggest effect on pandemic.
Mix?
I don't know.
I just put a big it's just as an example.
Because I've never heard of somebody getting a jar of mosquitoes with them before.
So I thought that was pretty eccentric for Bill Gates, so I put up that picture.
What does COVID look like in five years?
Emily, That's tough question.
I don't know we have to wait and see we can't even projects it's months now into the future in a meaningful way.
5-years is anybody's back at this time.
So yeah.
So over the years, the lessons learned, big lessons we have learned, we've been trying to fight mosquito based epidemics for over 50 years.
The lesson that we've learned as interventions are not effective once epidemic starts, so we need to do something before this epidemic becomes a massive scale and starts becoming a pandemic.
So the idea here is we need proactive intervention in the current that we want to be able to intervene with these interventions before it becomes a big problem.
And for that we need forecasting.
Here.
Forecasting is somewhat analogous to weather forecasting that we're trying to answer the big questions as is there an epidemic going to happen?
When what time are we expecting some of these big outbreaks to happen?
What types of vital streams or what types of diseases are we going to get?
And how can we contain this epidemic?
Can reduce a measures where if the original epidemic was pretty high, Can we do something to actually decrease the intensity?
Epidemic so that we can help and protect people.
And the idea here is somewhat similar to weather forecasting.
There.
If you know that is going to be a peak epidemic at some point in time, then you can go and actively advertised or spread that knowledge among the population.
Take protective measures and help mitigate that epidemic.
That means you're decreasing the peak, reducing the number of cases and many do that overall, the epidemic becomes much lighter and more manageable that's the idea is you give people the information so that they can protect themselves just like weather forecasting, right?
If you know going to be called today you take a jacket and make sure you put on your scars or glass to make sure you stay more same idea with this epidemic forecasting.
If we know that these problems are going to come, we can proactively protect people.
So here's a quick overview of our forecasting continuing methodology that I've been working with, with the different government agencies.
There are four key steps in the process that we have.
We're just going to do a high level overview here.
The first step is called modeling, where we try and model the fundamental mechanics.
Or the fundamental epidemics, epidemiology of these diseases we use what is known as a compartmental model but really what happens is in the back, there is a series of mathematical equations that are continuously being solved in order to understand how this epidemic progresses.
The math equation has, variables and constants so it does have some variables and some information this data is obtained from real world information and data that people are collecting.
So in our system, we model all of the humans that we would like to protect and to understand.
So we have the entire population of all 51 countries included in the model, including little islands.
So if you look at the small Caribbean islands, we even model those as well.
Then we have detailed models on weather and temperature because a lot of times the transmission and behavior of these viruses and people's movement is modulated by better and temperatures of the model incorporates those effects we have detailed models and mosquitoes if you're setting a mosquito based disease, we have detail models of mosquitoes we even model that mosquito life cycles like metamorphoses going from x to large.
Cuba to adult mosquitoes we even model the lifecycle of mosquitoes to be able to understand different intervention strategies that can be applied like larviciding, versus controlling adult mosquito populations by fogging and such.
And of course, we incorporate travel, so we have all of the travel data from all of the a 150 50 international airports and the different countries.
And the study the travel patterns between this and then the even have models where we include the crew ship travel, not just air travel, but also traveled or based on cruise ships and everything is included into this model.
So it becomes a pretty complex model if you think about all of this data.
Coming together but once we have this model, we then read the calibrated so calibration is the process where you use a machine learning technique there the computer kind of searches so you can see this black line.
That's the actual data that we have.
So that's the actual epidemic model.
That you see.
This is how the infections are going and then they'll flatten off.
So the computer searches, it tweaks the knobs and levers in this model.
So it's tricking them to see what best explains how can we adjust the parameters in our model so we can track the actual infection from that model so that process is called calibration.
You're trying to adjust or fine tune your model for this.
And we've performed all of this analysis on a supercomputer center.
Some of these run on a red-hot cluster on Canvas, we have a small supercomputer cluster on.
Campus.
A lot of the work, like the initial development and study happens on that.
And then for the large simulations that are run, they all run on the Ohio supercomputer center of Iraq, ran on the open clusters super computer cluster at the highest supercomputer center so we are able to complete about three months worth of compute so that means if you it had on it standard desktop computer that we have, it would take three months for this computation to finish, but now with the super computer, we can finish it in less than 12 hours so that's where the supercomputing comes into the picture where being able to take these large models and rather quickly.
Study and work with them.
So we can understand these epidemics better.
And then once we do that, then comes the forecasting part.
So that needs to be a fine tune the model.
Then we can see how the forecasting happens.
Imagine that the black line is not there, but it's there.
So that you guys can see and visualize.
And compare how this works.
Notice that forecasting doesn't give you a point estimate.
It's not like a point, but it gives you a range of values so what the forecasting basically says is on this day, if you were to look at someday, it could there are different possibilities, right?
The epidemic and progress in slightly different ways through slightly different.
Trajectories what you're seeing is that approximately this is how the epidemic is going to go so it can lie anywhere in that region and as you get closer and closer to the day of forecasts, it becomes tighter and tighter.
That means we are able to make better and better predictions.
And as we go further and further into the future our predictions will become.
a more and more approximate.
So that means we are projecting a year out into the future.
There's going to be a lot more variance.
That means the band over confidence is not going to be high.
It's going to be larger region.
But closer to today's settings.
It'll be tighter and tighter.
So that's how that forecasting.
All forecasting work this enduring weather forecasting works this way.
And once we do the forecasting this is where the public health experts and people who are making the supporting the decisions come into the picture what they do is they look at all of the different parameters in our model, can be control mosquito, eg.
Population.
Can we control mosquito bite rate?
To see if it will make an impact?
What will happen if we control international traffic.
Would, would that make a difference?
What would happen if we control people traveling?
If you have travel restrictions, what would happen?
So you can have all of these different kinds of questions that you can ask in the model.
And this model can come and tell you which what is the influence, what is going to be the impact of these kinds of policies that they can implement and by looking at the results of these policies, then the decision-makers now can come and say, if I were to do this in Puerto Rico which was one place where we had a lot of chickungunya Zika epidemic it looks like the initial infection is seeded is Barrett, will have the maximum issue.
So most of the work should happen in terms of screening people at airports, so that we try and combat that initial infection.
The time of that initial infection when it comes into Puerto Rico.
Because Puerto Rico is a big hub for all of the Caribbean traffics, a lot of people will go into Puerto Rico and then go to all the other Caribbean countries or South America.
So controlling that initial influx into Puerto Rico is the best bet.
And most of them.
Other approaches are not going to be as effective.
So this information then helps decision makers to try and contain or focus their efforts on that kind of a policy so that they can have the most effect or can effectively contained that epidemic that's the idea and of course, this is not a one time thing.
You can continously do is we do this cycle where you survey and assess, you predict.
I look at the future and then look at the results that are coming from this.
And then you use that data to inform policies.
The policies get implemented.
And then you start the cycle again.
Now with the policies in place, how are things changing?
And the cycle continues, right?
It's a continuous cycle.
And this is one of the nice things about epidemic forecasting is with weather, there's not a whole lot you can do to impact of other ready, that's going to.
Be called the only thing you can do is take some.
The extra clothes and keep yourself warm but with epidemics, you have the ability to go in and make an impact, make a change and that's why this is a continuous cycle where we have to the assessment and come back and look at how the effects are.
Here's a quick video.
This is how When I go to the Department of Homeland Security and such.
This is the kind of videos that we look at.
So let me see if I can skip into this video.
Okay.
So let me see if I jumped in the middle.
You get should be able to see how this you guys can see this epidemic moving.
And then we typically have these charts plotting.
Let me see if I can zoom in a little bit more.
So you guys can see how you guys can see this epidemic movement, blue areas showing light effects the yellow area showing higher infections.
And then you can look at them you look at these charts this whole thing is put on a huge display.
This is 8K resolution video and then it's put on a huge display.
And then we can see a lot more details when high resolution display.
And we can look at how the epidemic is progressing, how the infection is going in the different places.
So this gives people a more holistic, like a dashboard lock at the big picture, how this epidemic is progressing so they can try and understand the progression of the epidemic and make more informed decisions and policies.
Questions so far.
I know there was a lot of information to absorb, but this is what it takes.
A supercomputer churning out mass and processing massive data.
And giving out data.
Yeah.
Alright, we do have a few additional questions.
The first one is from Tony.
[SPEAKER] What data sources do you use and how do you retrieve the data?
Is that is a real time collection.
Good question, Tony.
The data I've been lucky in the sense that I've been collaborating with the Centers for Disease Control and the Pan American Health Organization, PAHO doors to organizations, what they do is they work with the local and the governments in the 51 countries and they aggregate the data.
They curate it And then I get access to the data that is a delay in that data.
So depending on the country, it can be anywhere from six to 8 weeks worth of delay.
So it's definitely not real, right?
So individual data collection happens that can afford up this entire chain, so it's a six to eight weeks bare minimum delay on all of this data that's the reality of the situation here.
[SPEAKER] And that's one is for Michael, what are the most difficult variables and assumptions to factor into forecasting?.
[SPEAKER] I'm going to talk about that next.
Case, reporting rate.
That guy gave me a lot of grief and I'm going to talk about it next.
[SPEAKER] Okay Great.
Another question from Pat, the inputs into your model seemed to be readily available.
Quantitative data how do you adjust the models for human variables such as science denial, apathy, etcetera.
[SPEAKER] That's a good question.
That's a great question we haven't gone that far yet, and that is exactly the value I get from these presentations, from talking to you folks, if you ask me those questions and makes me think.
Ooh, good question.
I don't have an answer for you now, but I will have an answer for you hopefully in a couple of years.
I've had my students work on that question, so thank you for that question.
[SPEAKER] Is from Elizabeth for academics such as you to consult, on government efforts, are there many other Miami professors doing similar work?.
[SPEAKER] Yes.
It is common for us to consult on government work, but there are hurt my faculty who are doing this too so that our Miami faculty, we even consult on legal cases so there are times where a court case might come.
And I know a colleague of mine, there the questioning some DNA evidence, and whether the DNA evidence is acceptable in a legal case.
And there are times where the experts work specifically on this technology.
That means they work like literally the entire 20 years of their career on that what technology they will go to court and they will they will assess the quality of the DNA evidence.
So that has happened too so it's not only with government, the even consult with legal cases as well.
So but I don't know the exact number, unfortunately, at the university level, I don't have a number.
I just know of several people who have done that in the past three thing and then the last question for now It's from Lucy the US share of is forecasting information with other countries which country or organization is doing the best job of forecasting?.
[SPEAKER] The CDC does share some of its forecasts, but at this point in time, most of this work that I'm doing is going into DARPA and some of the Department of Homeland Security.
So the things I'm sharing with you are the unclassified parts of this research.
I am unable to share the classified parts of this research.
But somebody told me this.
I could share it.
But then we wouldn't be able to let you go back home after Winter College, if you want to go back home.
I can share the classified parts of this.
So we still not completely naturalistic, confidently open yet but I'm hoping that will happen sometime soon.
Let's go to Michael's question.
Mary asked, what was the most challenging part of this?
Yeah the key challenge that I found was estimating what is known as the case reporting rate so this is what is how many people like than the countries collect this data.
And report it?
How much of the cases are actually reported, like how many people are actually being collected and they don't anyway.
With my approach that I have since I'm building this from a mathematical fundamental principles, philosophy approach, yes, it is hard building it that fundamental level is hard, but it's absolutely worth it because I can even figure out some of the missing variables in this equation and Estimated.
I made those values and fill them in.
So for example, this case reporting rate was one big challenge that we have, where interestingly, we found that there was about a 60% correlation with the economics of the place that means the richer countries.
Economically affluent ones seem to have a better case reporting rate.
That means.
Sick people are seeking more medical help and they are being tracked versus the poorer countries like for example, if you look at Haiti, Haiti was going through lots of after the hurricanes and calamities.
They were really struggling.
So people were being people were sick.
But they were not seeking actual medical help or that was not being tracked in a meaningful way.
So the percentage, like only 3% of the people were being reported and there was 1 so I was in the White House.
I want to set this was sometime mid October 2014, 2015, maybe 2016, something like that I was in