Navigating Complex Systems- Don Burke on Epidemiology, AI, and Modeling

John: [00:00:00] Hey there, welcome to the next episode of the flux, where we talk data decisions and stories of people asking what if questions to better understand the world. and create an intentional impact on the future. As always, we hope you can turn hindsight and lessons from our guests into foresight so you can uncover how you can make better decisions and create an intentional impact for others.

Today, we're on with Don Berg, co founder of Epistemics. So Don, why don't you give us a little bit of the story of what got you into modeling in complex systems, computational social science in the first place? 

Don: Yeah, thanks, John. I'm an infectious disease physician by training. and, early in my career went into the army and did infectious disease vaccine research.

 to do that, we did a lot of epidemiology about disease transmission. I, moved to Bangkok, and worked on tropical diseases in, Thailand, Japanese encephalitis, and dengue, and hepatitis, and really exciting stuff for a young guy. it, but it was, we, worked on vaccines. We developed a co development vaccine against Japanese [00:01:00] encephalitis that worked and saved probably 30 kids, from getting paralysis.

worked on dengue vaccine. Didn't really, get a successful vaccine during the time I worked on it, worked on hepatitis A, and handed that off to SmithKline. And I was the sixth person in the world to get the hepatitis A vaccine because we developed it when I was at Walter Reed. Yeah. So the point being is that I was a pretty applied researcher in infectious diseases and I really wasn't using modeling and simulation at all.

Yeah. 

John: You guys were just giving each other vaccines as you were developing them. 

Don: Yeah, that was the way you did it in the old days. If you believed in your vaccine, you were the first person to sign up for it. They said, today the ethics are that you're not supposed to do that, but to me that seems like a pretty clear ethical standard.

And then when HIV came along, we also did the same thing. Every time there was a new epidemic anywhere in the world, I traveled there, collected viruses, brought them back. We did genetic sequencing of the viruses. [00:02:00] Tried to make sense out of how the virus was evolving and transmitting. It was,we were the first serious global molecular epidemiology of HIV.

The reason the military was interested in that is we have troops all around the world and we were working on vaccines against HIV. And so we wanted to know if there were different types in different parts of the world, whether or not you'd need additional vaccines. So we're doing that and discovered that HIV was recombining, it was swapping genomes as well as mutating, and we can start to make sense out of, the, the molecular evolution.

So as I was studying that, I was wondering how the virus used its, what techniques it used to mutate and evolve.and read about a technique, that was being used in the early years of artificial intelligence, it's called genetic programming or evolutionary programming. And, and one of the [00:03:00] major methods for that, was a method called genetic algorithms.

And genetic algorithms were designed by a person named John Holland, a computer scientist at Michigan, who consciously borrowed from biology to say, how do, biological systems solve problems through evolution? And he applied that to computer science,and that was being used widely, to try to solve problems.

You will, you would write a, you would write code and you would allow it to become modified and you allow the. the modifications to recombine with each other,and at the, and then you would test which one solves the problem the best. And they and I came across people at the Navy Center for Applied Research and artificial Intelligence who were using genetic algorithms, who were, they were former students of John Hollands.

and, the Navy had invested a significant amount in methods for, machine learning, and evolutionary. [00:04:00] They were teaching torpedoes how to chase other torpedoes, and anticipating, what this, the strategies, the evasion strategies that a torpedo would use.and the, but the method, because it was originally derived from biology.

It was actually pretty close to the way HIV was evolving, and I had tons of data on HIV evolution globally, and they had a simulation tool for simulating, evolution. And so we started building, simulated viral evolution. And I cannot tell you how much fun that was. it was,my, my principal collaborator was John Grafenstedt, who became a long time friend and co founder, of the academics.

and so I was still in the Army at the time, and I would regularly go to the Navy Center for Applied Research, and we would have our,I would sit down, research sessions and ended up writing a couple papers where we simulated HIV evolution. It was really a cute system. We had a, we didn't code for, amino acids, [00:05:00] with,tRNAs.

We, we encoded the, letters in the, In the, Roman alphabet, A, B, C, D, you got the,and so the triplet codons that we had, coded for letters. And the problem that the viruses had was to solve how to spell words and sentences, and get that correctly. And so it was relatively easy for us humans mentally to follow that.

And we did, and So the first set of problems we gave them were to solve the problem of how do you spell core protein polymerase and envelope, which is a joke because that's what the biology is. And then we turned them loose, and then we let them have different mutation rates, their different recombination methods.

And we essentially had, contests between different evolutionary strategies, and it was really cool that, that it, we were able to, come up with methods that were pretty, the most efficient, and at the same time, they, and they turned out to be pretty similar to what HIV was doing in the first place.

So [00:06:00] that,that's, That was my epiphany. That was the notion that, oh my god, you can take electrons in code and, and that can simulate what goes on in biology and end up with a,a conclusion, where you didn't study the biology, you studied the simulation, and came up with, what was the same sort of read.

You've got a better understanding. of the system.and that was when I became a born again modeler. That was my life changing moment. that after a long career, a successful career in building real world stuff, that I realized that I could do that better with, if I used modeling and simulation. so that, that set me off on my modeling and simulation career.

John: So from that moment, what came next? 

Don: Yeah. so then it was just, that was just about the same time that I, retired from the military and went to Johns Hopkins, took a professorship there and director of the center [00:07:00] for immunization research. So I kept doing pretty much what I was already skilled at, but I started to build my capability, my own personal capability in modeling and simulation.

I don't have any training in that. I don't have a background in that. So I had to do on the fly. the, training way I did that was I got some graduate students who were interested in modeling in real world. I knew, Dengue and, I knew the real world data inside now. And so I could help the students by getting the right data to feed the models and ask the right questions.

And I ended up having a couple of terrific graduate students who taught me how to model. And it was a gradual process. So Justin Lessler and Derek Cummings, they're both professors now. Derek's back at Johns Hopkins as a professor. Justin's at North Carolina as a professor. And they're, they're really bright guys who were just, we had fun too.

So that was the next process. [00:08:00] After that, once we got started, I was able to get some smaller grants from, that related to, climate and infectious diseases of dengue and others from, NOAA and the like, and that worked out okay. And so we started to build a little group around modeling. And, 9 11 hit, and that was when, that was a major inflection point in modeling in the United States, is that, people took, started to take modeling and simulation more seriously.

After 9 11, there was, the anthrax attacks at the Capitol, and, then there was serious, so there was serious concern about biodefense and biological warfare, and particularly the concern was about smallpox and whether or not smallpox can be used as a biological weapon or not.and and this is a hot topic.

Should we immunize everybody in the United States against smallpox? And at the time, under George Bush and Cheney's White House, the [00:09:00] predominant opinion was, yes, we should do that. Because the threat is so great.and there was a good guy at Yale, Ed Kaplan, who was in the School of Management, wrote a paper in PNAS that said, Yeah, of the options, that was the best option.

And it was a modeling and simulation paper. Not agent based, but equation based. And the people who were real world experienced in smallpox, particularly D. A. Henderson, who had led the group. global smallpox eradication, were appalled at the notion of immunizing everybody with a vaccine that had a lot of complications.

And it, as it turned out, DA was my suite mate, at Hopkins, so we got to know each other quite well. And at that time I, I had, gone to a meeting and sat next to a person who was an agent based modeler, Josh Epstein, Josh and I hit it off.and, decided that we could probably do some modeling together.

Uh,so [00:10:00] I,now I was at, At Hopping time, we'd go down to Brookings where he was, regularly and we would meet and work on agent based modeling. Josh, was on the program earlier, but he's the, one of the godfathers of the, of agent based modeling for decision making in human societies, growing artificial societies and the like.

and, And so we built some smallpox models and we're interested in the effect of vaccination and what could we do ring immunization versus immunizing everybody and just as we were doing that, the DA,was at the White House and then became the assistant secretary, for preparedness and response.

and a major physician at the time was worrying about biohazards. And he was adamantly opposed to the idea of vaccinating everybody in the United States when they eradicated smallpox globally. They didn't do that. They went around and did, every time they found a case, they would do ring immunization.

They would vaccinate just the people in the area [00:11:00] around a new case. He thought that, if we had at least a smallpox attack, that's the way we could contain it. But we had this, a Yale professor's paper, he had Dick Cheney, he had the opposite side of people were saying that's, that they should go ahead.

and, So he asked if we could do some models as well to compare, the different strategies. there were a couple other groups that he invited in as well. One was, Steve Eubank at, Los Alamos. Another one was Betz Halloran and Ira Longini from,from Emory. and, we met at the, Stonehouse at NIH, at his, DA's invitation.

He was able to cut us some fast contracts to do the work, as an, under an emergency basis. And we started, we had a work group on small ox. And this is the proto mitis. Let's just say, yeah. Okay. and so we, yeah, we did our modeling and, you won't be surprised to find out that we all modeled and agreed that, that you didn't need to [00:12:00] vaccinate everybody in the United States.

Yeah. And that now we had dueling models about what the right answer was. Yeah. And,So that's how we got started in the agent based modeling and the modeling for infectious diseases, but also it was the realization that if you're going to be making important decisions about big strategy like dehumanize everybody in the United States.

having, I think it was Bernoulli who said that having a little bit of, modeling in support of the decision is problematic at that. And,so that's what got us launched on the, and then what, then the NIH did based on the, what was perceived as the value and success of that.

group, the NIGMS then did institute the MIDAS program, the Models of Infectious Disease Agents Studies. the, and we, then we successfully, my group at Hopkins put in for one of those modeling groups and won it. So that was my [00:13:00] beginning as being a modeler. I'm the first person to confess that I am not trained as a computationalist, I'm trained as an infectious disease epidemiologist.

My main contribution all along has been to understand the problems, pose them well, understand, find smart people who can work on them, and bring resources together to make that happen. And, my, one of my You know, my heroes in analogy,is Barbara McClintock, who worked, won a Nobel Prize.

for the, genetics of corn, of maize, and how there were, transposable elements that were not, it wasn't strict Mendelian, that there were some genetic elements that could move, and she showed that, and, but her biography is entitled, a feeling for the organism. And I always thought that was such an apt way of talking about that kind of expertise that, with a deep understanding of a particular problem that gives you the sort of the, you know, the [00:14:00] leg up on somebody who is, doesn't have that.

And so I think I, I'll modestly say that I think I have a feeling for the organism of having a full career working around the globe on infectious disease epidemiology. I think that's one of the things that I, that lets me see the right problems and pose them well. So then, and then from then on, the MIDAS program at Hopkins.

And when I moved to Pitt, I brought that center and we became a center of excellence with the even larger group. And had subs in, with Josh State as a sub. Roy Anderson and Neil Ferguson and subs to us. And,so it was just, it was a really, it was probably the, one of the most powerful modeling groups in 

John: the world.

So Don, from the beginning of Midas to where you've seen Midas and at least agent based modeling and simulation in the infectious disease world evolve over the last 15 or so [00:15:00] years, where are you excited about? New areas for agent based modeling and infectious disease simulation to go. 

Don: the, I will continue with that Midas story a little bit more to fill it out that, when we applied for a, center of excellence, I did ask Roy Anderson, who was, probably Roy Anderson,and Robert May were the, in England.

Where the leading modelers, they wrote the book that everybody used on modeling,and non U. S. were not allowed to compete for the NIH funding, so they were happy to be a sub to us, so I was quick to corner some of the best talent in the world. And, and among them was, Neil Ferguson, who had been a, postdoc, with Roy.

I had met him some years earlier, when he was a postdoc. And,and, very smart guy, trained as a, a, statistical physicist at all, right, computational tools and understanding of complex system dynamics.and so we decided at that time avian [00:16:00] flu was a threat in Southeast Asia. The virus was raging through all of the poultry in Southeast Asia and there were a few cases of Human cases, and so the concern then was that there would be an outbreak, and a major epidemic that would come out.

And, and so we, built some large scale agent based models of, transmission of influenza. And again, because the MIDAS group was doing this, Steve Eubank and Betts and Ira and others started to work together. So it was a collective. And at the time, again, the White House, now that there was some credibility to the process, Richard Hatchett was at the White House and he convened the groups to advise them about how to go about, what are the best strategies for containing a major new epidemic that led to the.

The targeted layer of containment and the like that was subsequently used. [00:17:00] so those papers that we wrote, Neil and, and I and a team, were one on, for Southeast Asia, one for the United States and the UK, where they were published in Nature in, 2005, 2006,were influential in terms of how you approach, your, influenza epidemic control.

I should add as a interesting parenthesis is that initial model that modeling platform that Neil built for these models was what he used when COVID broke out in 2000. So he had a platform that was already ready to do the simulations and simulated COVID, for the U. S. and the U. K., because that was a whole lot different than what we'd already done for influenza.

and, and correctly said, that the epidemic, if given the trajectory, could kill a million Americans and half a million people in the U. K. very influential paper, but [00:18:00] it was the son of, but the problem with that model is it was a, it was, it, nobody else could use it other than Neal.

it was a research hack. and,it was good, but it wasn't, final. So that, that was, so there were agent based models out there that were doing well. but, when I went to Pitt, I, hired, John Grefenstead. I called him up and said, John is a friend, and asked him if he would be willing to come to Pittsburgh to lead a center for infectious disease modeling in the public health dynamics laboratory.

And I didn't know it at the time, but he was very interested. John was born and brought up in Pittsburgh. His parents were there. He had, I think he has 10 siblings or something like that are scattered across an amazing family, by the way, every one of them, judges and musicians and,business people, just extraordinarily talented family.

and John was happy to come back. And, so we worked together and among the things that we did was to say that we really needed a, an agent based modeling [00:19:00] platform that was usable by not just, the creator, that it needed to be done by industry standards. and so we had a team, Bill, start to build that under John's direction.

Yeah. And that was what the original FRED model was. so there have been a series of, first, from, this is my vantage,everybody has their own perspective. I first worked with, Josh in getting the toy models on smallpox, and then worked with Neil on getting the, research, software,for real world flu problems, and then, worked with John to start to build the platform that would be, a professional grade software platform that, with realistic populations and everything that FRED has.

So that was the progression from my point, and my job on this site was just to be, I was the, the program director and, and Help to guide the thinking of the people who are working with me that these were important [00:20:00] problems to work on. 

John: as the next evolution of FRED comes to be on the more commercial side and working on the getting FRED and agent based modeling adopted more and more, where do you think the field is ultimately going to end up going?

Is it going to be just about disease epidemics or are there other areas that you think, agent based simulation is going to, you 

Don: Yeah,for the entire time that we were building the FRED platform, then, John and I talked about this, John G. John Gruffenstedt and I, talked about this many times about, that It shouldn't be just a platform for infectious diseases, it should be a platform for transmissibility of anything which is transmissible human to human.

and for, whether that's,attitudes, beliefs, behaviors. Whether or not it's, political persuasion, whether it's anti vaccine movement sentiments, whether it's [00:21:00] religion, political affiliation. So FRED has that capability of representing anything which is transmittable,social contagions as well as,microbial contagions.

and, and the reason for that is that infectious diseases are one very important real world practical problem. But so much of the, dynamics, social dynamics are currently, not modelable. that, it's a, my own perception, I'm not a social scientist, but I do observe that, when we try to model infectious disease epidemics.

Modeling the, the virus is the easy part. Modeling the humans is the hard part. and, I think the field will go is that we will get better at representing how, decisions are transmitted through people to each other. And then, and that the vehicle for transmission can be in person, can be social networks, can be, and there may be [00:22:00] features that of transmissibility that favor one, route versus another, but, I do think that, the, that where the field of modeling and particularly agent based modeling will be going is how, attitudes and beliefs, and, behaviors, emerge, how they're transmitted, yeah, and to do that we're going to have to get better at understanding the mechanisms and the data streams.

Those are tall, those are big lifts, because we don't, we're actually pretty lousy, we meaning the entire scientific community, pretty lousy at both of representing mechanistically how people, Make decisions. . And we also don't have data streams that are con, that conveniently feed those, mechanistic representations that we already do have,in health behavior and other areas.

We, we have a lot of. simple theories about how you do that. But, there are things like perceptions of risk, [00:23:00] perceptions of self efficacy, trust, those are factors that go in. But then how do you measure those things? Where do they come from, if you want a real world? Nice thing about Brad and I, and with these, the real synthetic populations, What I can say, trust in sources of information will vary widely depending on who you are and where you are, in your own,genetics, your own family, your own history of the region, and, so even a single factor only got to know, it's going to change depending on who you are.

The individual agent, and in fact we can do that, and the assumption will be that for other factors like whether or not you have a belief that you, that what you do can actually change things, that probably changes from person to person and based on their characteristics too. in the long run, I think that's what we're going to do.

You [00:24:00] know what the fantasy end of the day is. Let's hear it. Yeah. We will be doing brain scanning that allows you, where there are, where these different, where the, the trust area lights up in the brain, and you'll be able to measure that. And instead of just asking people, do you trust this source of information, you can show them a picture of huckle slime and they will either, glow red or glow blue, and, and that will, give you information about that particular.

Aspect of the what's going on inside their head,you and I have talked about this notion of little minds is as it is, we have little immune systems in our models right now, we do represent, immunity instilled in the individual agents and how long it lasts and what are the factors that, and so we not only need that, that's the nice thing about an agent based model.

You can put in and. a little immune system, or a little mind, or a little spleen, and [00:25:00] represent that. Little spleens probably aren't quite as important for, direct contagion, but little minds are terrifically important. So I think that's, where it's going to go. The,there, I haven't, I can't chart out a route there, but I would love to get social scientists together.

To discuss if we want to move the field, to understanding the dynamics of contagion, behavioral contagion,what do we really need to do? What if we had a moonshot for understanding human behavioral,and by the way, this is, I want to say moonshot, This is such an important issue.

You ground our epidemic control efforts to a halt in some parts of the country because of this irrational fear of vaccines and the behavioral contagion that was not based on rational decision making. So understanding that, if I were King,this would be something I would be [00:26:00] investing in, I would be, And there, there is other, are investments in, in the, the ASF and others do have some efforts in this area, but I, I think that it's under, under, Underfunded and not well organized.

and what I believe is that agent based platforms like FRED are going to be instrumental in helping those fields advance. 

John: Absolutely. So even looking at fields of economics and other things where they assume, for the most part, rational decision making and agents, the COVID pandemic and plenty of other examples illustrate that's not totally how people make decisions.

And, And then that's something that's a good virtue of agent based modeling. One of the things that we've talked about in the past is the idea of the campfire concept, being able to bring different types of people together, the computationalists, the qualitative type researchers. I think that might be another good thing to discuss, [00:27:00] and the grand vision of being able to communicate across these different fields and different types of people.

So it, I think it'd be good for some of our listeners to hear your grand vision on the campfire concept and how that plays out. 

Don: when I started to get into modeling, more than 20 years ago now, I,I had my students read a book called Serious Play. and it was by a person at the MIT Media Lab, Michael.

I think the,and, in the whole idea of his notion was that serious play was that By having a thing in front of you that, you manipulate, and you noodle around with it, jointly, you, forces you to share explicit representations of what's going on and whether or not that's a, an Excel spreadsheet or a clay model of the latest Chevrolet or, any kind of model that is built, that is, isn't in words.

But is yet [00:28:00] another representation of the ideas, in a different medium, is, it allows for communication. And, so that, we, so John Reffensen and I, when we, wrote a little paper on this,we, And then, a campfire, that, where you sitting around a campfire and the model itself is a place where you swap your ideas about what's going on by having the explicit representation in front of you, then you, that's what you end up sharing with each other.

But when you use words or even when you use equations that, that lump, lots of sub variables under a variable. It can be easy to mislead yourself about what exactly you're talking about. And one of the, one of the, virtues of an agent based model, it forces you to be explicit. huh. On not only what's going on inside the little mind or the individual, but also the [00:29:00] processes that, that, allow the transmissibility.

Yeah, I, I, yeah, so there's more to modeling than just modeling for an answer. Yeah. Then, the,it's the, it's, the, this is the colon, the, medium is the message, that, the agent based model is the message, that,and so it's a way. Now, that's, That only works if the people who are the decision makers actually are sitting around the campfire.

and if you, you sit there and nod and you come up with an answer and you take it away from the campfire and you just put it on their lap,they're not going to have that same, Feeling of understanding or ownership or engagement that, being involved in the processes. So that's one of the challenges in modeling is that how do you make the model be the model of the decision maker, not your model?

Yeah. But yeah, that's a valuable notion that campfire or the, the series play. Yeah. notion of, of [00:30:00] models, and it's a tool for shared thinking. 

John: So I'm going to take us to getting to our what the flux moment and talking about that from a decision making standpoint. But, one of the people you mentioned earlier on the podcast and D.

A. Henderson, I think it might be good when you look at the impact that he had on smallpox. He talked about,the difficult thing about that and how long that took and the ups and downs of, Of doing that. And you had a pretty close relationship with them. we've talked about the challenge of getting agent based models adopted, getting this campfire into decision making approach adopted.

I want to open it up to you a little bit to talk about, some of those lessons from DA Henderson on, this is like a life's long bit of work To see come to fruition. So I'll just give you the floor to, to talk about that before we jump into decisions and some what, the flux type moments.

Don: Yeah. So I got, hey, DA's, one of my heroes. people who've actually made a difference in, [00:31:00] in the, in, in our planet. In the human condition. Yeah. he's one, he's right up there. the,he also happens to be a Donald. from Cleveland, he was a professor at Hopkins, he was a little overweight, and, so it was, yeah, so D.

A. used to come into, the office wearing a sweater vest, and one day I commented on him, he said, yeah, you wear it to camouflage his gut, and I, and so I, from that point on, I started wearing it to the, to the,the, DA wrote a book, called Smallpox, Death of a Disease, and, he asked me to write, something for the Dust Jacket, which I did.

and it's a, his view of what it took to eradicate smallpox, and it's a,It's a testimony to the, the persistence that he and the others, teams had to go through. They, the political obstacles, this, the, the, the, the physical catastrophes, whether it was earthquakes or, hurricanes, the,the political, fights, with, in the war zone.

And just, [00:32:00] and the number of times where they thought that all had been lost, that they had lost the war, and yet they persisted. and, so as I said on my dust jacket, I said, this is the stuff of public health. it's the grit, it's the, the digging to get it done. And I found that to be.

I think you can apply that same lesson to anything we're doing, is there going to be times it's not going to be clear sailing, and, so I, we've been at this for a while. I, and, I won't be surprised if one of these days it is going to, there will be, the inflection point.where it will just, it'll take off.

we're not there yet, but, hopefully we will be. And I should, then going back to small box him, they, the eradication of smallpox, has paid for itself, every 35 days, the last 20 years, and they,those are the kinds of things that, that, that do matter.

John: So moving us back into modeling and decision making.whether it was military [00:33:00] career, time at Hopkins, time at the University of Pittsburgh, other things you were involved with, where have you seen a situation where If you were able to go back in time, being able to use the tools that exist today to model things out, to run some simulations, to improve a decision that was made, can you think back to one where you're like, you know what the flux were we doing at that moment?

we should have been able to run some simulations there to, have a more clear idea about what decision is the right one for us to take. 

Don: as I said, I had a pretty clear demarcation between my. pre modeling life and my post modeling, my current modeling life, and I, so the question is, are there things that we can do, we should have done, in, say, with the dengue and Japanese encephalitis and hepatitis A.

And at that point, we were doing vaccine trials,but we, we didn't simulate, we calculated, its expected efficacy and we [00:34:00] calculated population size needed to, achieve the likely efficacy. So I think it would have been very useful to go back on those areas and simulate the processes.

I've worked in lots of diseases where we had lots of data. We could have simulated those in a better understanding and made better decisions about control. Yeah, I think that's probably the main area. Yeah, as I told you, it's an I tipped my hand on this. I already gave you my epiphany, moment.

And hard to find a pre epiphany, But those would be it. Anybody who works in field epidemiology should be using this kind of simulation to guide their thinking. 

John: So don't have two more, two more questions left. one is about how people get into modeling and what some [00:35:00] of their backgrounds are.

you're coming at this from having gone deep in applied epidemiology and then having your born again modeling moment. There's others that come into modeling and simulation from computer science. So when you think of teams that you've built in the past, others who are good modelers, but also good modelers who can translate.

To decision makers, what are some of the backgrounds of people that you found consistently lead to successful modeling and simulation efforts? 

Don: so when I, John Griffiths that I first started to work at, he was a computer scientist and I was a person doing molecular epidemiology of HIV. We had very little common vocabulary.

the, he taught me, When I taught him how to spell DNA, that, that we had,yeah, we had to just learn each, a language together. And one of the problems is it's rare to find people who are conversant in both fields. It's, not conversant, but are expert in both [00:36:00] fields, what you want is somebody who's really, has expertise, and, and as I said, I feel, I Comfortable in my skin that I, I have expertise in, I don't know that I have, I know that I don't have expertise in, in coding and modeling and the like,and I have some knowledge of it, but I, I'm not an expert in that.

So I'm one person who was brought over from the, the applied real world side and there are others. and when you teach this kind of things, it's really hard. Some places have done it better than others, but the historical, the pedagogical stovepipes have been, you learn infectious disease, you learn public health, you learn computer science, you learn biostatistics.

the whole system in our field was set up,to be stovepiped, all the way from the NIH, funding sections on down. And even [00:37:00] today, there's been a lot of realignment and a lot of departments are now getting. about teaching people to be at this interface, much more so in the area of the biological sciences, which are probably 20 years ahead of the social sciences in terms of, youcultivating a, a, a rising,set of young scholars who are competent in both, computational biology.

There's less in the emphasis on computational social sciences. Part of the reason for that is the computational biology was driven by The, it was almost driven by the need for computationalists with the streams of sequence data and, the nature of the biology is strings of things, and you needed the computationalists to make sense out of those.

In the social sciences, we don't have streams of things, and, so it's a little harder [00:38:00] to, to, that the, we're behind there, but the data is getting richer and richer in terms of being able to have, find automated data, but in social sciences, it was, that's the way the data came out, and,um,so I, as a dean, I was, I, and director of a center for modeling, I was, And people who are smart computationalists to teach, to learn epidemiology and find smart epidemiologists, to learn some.

and there were a few that they were able to do both,and they go on to be, the leaders and the professors. But it's a, but it's a, that's a big lift, to have somebody be too true. not just interdisciplinary, but, too transdisciplinary.

I don't know if that answers your question exactly, but I don't think we've got it right yet. There are more programs for computational epidemiologists, but that's different than [00:39:00] classical statistical epidemiology.and then, we just need, and when I talked to you earlier about, making investments in the fields, this is where, one of the areas you need to make investments is teaching, these, these.

John: Transdisciplinary 

Don: means 

John: sit on last, last question and, for the listeners know that Don's going to probably be on one or two more of these podcasts in the future, but what's one of the things you're working on right now that you're really excited about? 

Don: I'm, I'm 78 years old now.

I have retired. I, have had a, a long and exciting career. And,and I have watched other. Senior, scholars, try to stay at it too long.that, that it's hard to stay current and start to stay on top of things, in the scientific fields. But what I can do, is I can be reflective about the history, of what I have observed and what are the lessons, that I want to see passed on.[00:40:00] 

And so that's what I'm doing. I, it's been my hobby my whole, life. Anyway, I was the archivist for the American Society of Tropical Medicine. I've written a number of articles over 

John: the years on You have a small collection of, books and journals, from some other famous infectious disease researchers.

Don: And I collect them. I also collect in a library. it's natural for me to do that. I can feel good about, this is something that few people can do successfully because I'm looking back now with it. So I have written a couple of articles already. I wrote one on It just came out on the origins of the 1977 H1N1 epidemic as being triggered by a non transmitted small outbreak of swine flu in 76 to what was seen as a possible swine H1 epidemic.

[00:41:00] epidemic turned into a real H1 when people went back and went into their freezers and resurrected a, what was already an extinct virus and that became loose on the planet. and so it was a self fulfilling prophecy epidemic. It's the, we were so worried about it happening that we made it happen.

And so I think there are some lessons. For us in knowing that, that story well, as well, the,I've written some on, how, the, we, we talk about,SEIR models, as one of the main forms of the equation based models, but in fact, the E in SEIR refers to exposed, but the way everybody uses it, it's really people who are not just exposed, but already latent and infected.

and so I was curious about how that, that mis connection of the terminology got built in, and so I went back and went through all the papers, and again, what I try to do is go back through the papers, and if I know, and many of the times I [00:42:00] know the individual that was involved, and so I'll call them up and we'll converse about it.

So on the, H1N1 epidemic, I, I talked to Peter Pellissier, the virologist, and Frank Topp, who is an epidemiologist at Walter Reed, for this paper on the SDR model. We talked to Herb Ethcoat, and,and,yeah, and, yeah, just said that it was just, it was just a accidental introduction that got carried along, it was a Bounder effect, and, and so many of the things that we do in academia, yeah, are, just be, they get carried along,For reasons that are not the best of reasons other than convenience.

So that's the kind of thing. In the past, and the major project I'm working on now is that our school, School of Public Health got a major gift of all of John Ossock's original laboratory equipment and his laboratory records, including the records from 7, 500 school kids who [00:43:00] were in his trials in Pittsburgh earlier.

and so I've been going through all of his papers and going through his,publications to reconstruct exactly what he did and what were the other problems that he overcame. So I'm not going to try to write a biography. There have been several good biographies about Selleck, but what I am going to try to write about exactly what was the science of what he did, because all the biographies that are written about him are pretty good.

Biographies, they are not, they're not analyses of his science. . And I think I can do that. And I, and also in large part, 'cause I have access to stuff that isn't in the publications. and, so I, that's probably taking most of my time these days. But all of the common theme here is, is, looking for lessons in history and codifying those for, for the, for people in the future.