[00:00:05]
>> Ok well. Thanks everybody for for sticking with us here and we are absolutely delighted and thrilled to welcome our. First Speaker today Brian Greene South grand fellow is professor of ecology and evolutionary biology and public affairs at Princeton University in New Jersey he's a population biologist and studies processes that occur in populations at different scales and how infections move through such groups of organisms his work is crucial in helping to control disease in humans and animals his work is theoretical as well as based on large data sets demonstrating how the density of a population randomness interacted change the size and composition of populations and alongside colleagues from the National University of Singapore he studied measles in developed countries and is now extending those investigations to hooping cough and other infectious diseases amongst other honors brined was awarded the t.h. Huxley medal from Imperial College London in 1901 and the scientific medal of the Zoological Society of London in 1905 and today he's going to be talking about what cross scale research can tell us about predicting understanding and mitigating future pandemics I couldn't think of a more appropriate talk for our symposium Brian please take it I thank you so much.

[00:01:23]
Ok so. In that initial time I'd like to rush through and talk about. The complexity of academics in this audience is very expert on not and might have given great introduction yesterday I'm sometimes things turn out to be dynamically simpler I'm not and I want to. I'm a traditionalist in this way I want to I want to advance the idea that we should always compasses more complex models and this audience is right very up I know it's nice to get it and then want to go on and talk about use not to go on and talk about pond make an epidemic down on mix across scales and how we deal it modeling with particularly in an odd logical an evolutionary complex taste and then I want to spend a chunk of time just on my own thoughts about research and global health priorities and they'll be some of a lot about this but I'm sure you'll come up with better ideas than I will that this is I'm going to touch on work that's covered many decades a lot of great collaborates as an unkind fund that's so like many many people on this on this you can call we dropped most things in the last year and and this is many my colleague just came back off and I think Princeton and I grouped and and and modeled lots of stuff from the local dynamics potentially on the Princeton compass to.

[00:02:46]
Him and logical and uncertainty the seasonality on and some quite applied things locally but the day job really for me for a long time now has been looking at the population dynamics of endemic infectious diseases and yes Mike yes my favorite which is measles and you see that that appeared to be relatively simple patents you know this is the pre-bought submission era and you can see that there was an epidemic every other year more or less the more on the role of the stock this is one thing this is 1950 now.

[00:03:17]
Can we make simple models of their spots that you know have some temerity I'm thinking you can because we know when markets summarize this brilliantly. That there's there's a ton of that which I mix boiling away underneath there are individual had to join a tease and because it is a predator of memory cells in the main system is beautiful intricate human logical dynamics going on and there are there are experts on all these scales all McColl which is great it's great to have breath of expertise.

[00:03:48]
I'm going to I'm going to try not to get that with the become sometimes make simple models because music is unusual it's it this would not be true of carving $1000.00 or more influenza you read it you don't shed the virus again once you've been not surely infected then the vaccines are pretty excellent as well so the amenities very strong.

[00:04:07]
And it's very very transmissible and this is a rewiring randomly rewiring network maybe a childhood infection and therefore you can sort of close one eye and hope it's in my section at some level but you know it's always really important to test against or intricate models on a crucial one really is that these epidemics they didn't do anything anything particularly strong to control them they just they just the same dip is going to happen and therefore there isn't big behavior change in behavior change completely calm complicates things obviously I'm not Sydney been to non-pharmaceutical interventions in the pond that **** and again this audience will be very expert on this.

[00:04:51]
Epidemics depend on the reproduction ratio now in very common parlance locked in the name in one thing infection susceptible and infected recovered tends to deplete susceptible is that for the effective rate ratio goes down when it passes through when the epidemic drops so I'm going to persist in calling this the epidemic clockwork and this this idea of herd immunity which which you can see strongly in measles and measles epidemics and measles vaccination for example so what's what's the kind of basic dynamical Ponson And again just ish and then one can obviously make very simple my selection models and just to show these it's a very very simple 2 dimensional system really with an extra Kika in that those seasonal transmission is obviously stuck us to city and on seasonal though it's a simple clockwork potentially seasonality and struck us this city can introduce lots of dining out complexities.

[00:05:48]
Chaos in particular so so we can we can take there are great historical records of missiles I can point to it's dates and we put on the web recently but you see that we can make models and run them folded and fit them to the shelter and the state space models and they capture this lovely bifurcation from biennial dynamics in this case and that's because of the baby boom high birth rate the epidemics the susceptible to get chopped up quick us and they are there the dynamics are.

[00:06:19]
Around it all and then it widens out into this biennial parts and so this is just a way of saying that there's some image and simplicity yes sometimes under those those. Here a stick that I gave at the stock you might see simple dynamics. Too we can use these simple noddles and it's all it's really great to have a hierarchy of models where you come back at simple and complex models what's the what's the mom if the old is that I'm on a field of simpler dynamics in a complex model and again you fix an expert on this this is lovely work by Rachel Baker.

[00:06:52]
Earlier in the epidemic there was a lot of discussion of seasonality of covert set me the. Endemic around a virus is a certainly when to diseases they like low to mid t. But but Rachel showed nicely I think that because you have such violent epidemics strong susceptible supply at the start then you wouldn't expect to early on at least see the effects of seasonality in a subsequent paper later on and I know you do tend start to see the effects but the susceptible dynamics would dominate so that was kind of a use of a simple model because there are then lots of complexities.

[00:07:32]
The dynamics is a kind of Ok in the face of these you can you can get away with simple models to some extent if you're culturist but there is super spread as a space many models for spatial dynamics the cell data records functions. And I know the things that exploded but they were already revolutionary before the code upon the mc they've really exploded loaded then and no pharmaceutical intervention social forcing much squaring and so on I just want to touch on stone some and some complexities we looked at which make things even more unset and getting some more dates and know that the nature of immunity against going to $1000.00 was not clear clinical It was a transmission blocking pathogen host of elation a man pathogen community dynamics which I'll touch on at the end so this is a lovely paper by shiny side Raleigh arms cut on lead by shiny side I'm kind of like watching that and they made a very simple model where you can you have to stop the strike you have secondary transmission and you have the strength of retransmission in the strength of secondary susceptibility and therefore you can titrate these to go all the way from the simplest silo to that to the s.r.s. model where you go back to complete susceptibility and obviously the simple model is is more optimistic in terms of the epidemic dynamics and that's all they were able to look at the week in 80 case where you got recurrent epidemics and a pessimistic case on a strong immunity case and the bottom row here is is deployment of the vaccine and you see that there is again a much more optimistic case year in terms of the original strain that the vaccine sic to be doing well against that so so so we're in the optimistic case however obviously.

[00:09:27]
New strains of the virus. Potentially actually to some extent but also potentially worrying and that brings us to one of the evolutionary implications of reinfection in particular this purple area here and you can read the paper in Science to look at the purple area here a 2nd degree susceptible So how important are they in terms of evolutionary going on and this recalls previous Where could be done and this is comparing measles and influenza just an accountant both are in a virus is both imperfect or a correction on reproduction but it's amazing that the strains different strains different variants of the virus the different different ice year but the Cross a bt is so strong you get out like dynamics and you're on the.

[00:10:21]
The surface Moloch in phylogeny is like a pumping phylogeny he's done time strong Patton's and influenza because because and the bio mantle and biology biological difference is really interesting he won't get through it because you can get reinfected in retransmit particularly when there are a cluster jumps for example you get this lovely lot alike phylogeny and you've got strong evolution of the skate and in this this paper we had a very simple framework with if you had a meaning pressure then if it made precious very strong they'll be a lot of strength of selection and pole tennis at a brilliant.

[00:10:58]
Stage for this yesterday but but very low viral abundance and if I were abundance to some extent controls transmission un like but you won't transmit and it's the other way around here there's not much selection and strong transmission and probably where influenza or maybe coated 19 are is that some intermediate level where you would have strong transmission of genic novelty So Shadi and and.

[00:11:25]
And Caroline to put this into a paper this is in medical leave and you can look it up but you. Looking at secondary infective as the kind of denominator of difference all if you had one of the votes and you had to use I'm not sure infection and titrating.

[00:11:42]
What the what the things that the threats and this is just for one city but obviously be very spread. In different regions what the threat of potential evolution is the bottom line of this paper really is that the this is there's obviously a moral case here is why other than epidemiological One is we need to vaccinate the globe we need to knock infection down everywhere to minimize new variants so so.

[00:12:08]
This figure I love I love I love this sort of figure and it's been great to see many versions of it on the front page of the symposium on impulse and his talk yesterday and this is for this is for influenza thinking about the different scales from the molecular level up to the global level and the father genetic level and there's there's a lot of modeling work at the population dynamics scale as there is there are wonderful folks within whose modeling folks who work at this scale a real gap for me is how we quantify transmission is really hard to do and I think that the sustenance sessions going to be very interesting to look at in the sense so just to bring this old all this together that let me tell you some personal just just syncretic research priorities pathogen surveillance is obviously key and this is really really a global health priority public health and genomics surveillance one health surveillance.

[00:13:08]
Also very valuable for vaccine design syndrome x. a male and a really interesting cross scale frontier and this is a paper by James Hay and and Mike Miller and colleagues some from Hollywood is is is a sort of a cross tale thing where you take the cycle fresh of the city Valley from p.c.l. from pathogens Lehmann's and you can read this yourselves back changes this is in a model but they show it in data but changes over the course of the epidemic you can get a feel for where you are in the epidemic trajectory from the c.t. valleys as well as from the epidemic pots and I think that's extremely exciting read the paper I was very struck by that that's really the start of a new frontier and then the frontier I think is we we we trace a nation an epidemic dynamics often just from the cases but if you talk to the meteorologists they measure all the variables of course with satellites another and another measurements and with an imminent logical observatory and this is this is again by Mike and championed by humans.

[00:14:08]
Just make up and others of us if you if from exhaustive search all the jew run the world you could measure the susceptibility in infected individuals you're in a sense not doing an end step and prediction you doing a one step production there's a ton of other good reasons to measure susceptibility on immunity as well.

[00:14:27]
I never attend about the things that you want to measure obviously and some things. The key thing and this is a n.s.f. thing and the pen making natural and social sciences extremely important I think I think that's part of the observatory system's analogy and population level of men it's extremely important and as I said transmission biology and the but seasonality part of the winter viruses that we read by Jeff shaman was just by matching patterns in a population that the Democrats too and some experimental infections of animals to things like immunity there's been an explosion now of looking at for asperity viruses at the very dynamic aspects droplet chemistry all kinds of things and calibrating that against human movement is a really important front I think and post pathogen evolutionary biology those cross those dynamics very hard to do but extremely important I'm not going to touch on it but social dynamics and how you quantify it the mask wearing social dynamics folks it has its incidents along and in terms of models and competition model again great lessons fine and in our area I think by Jeff shaman another cat Shay and other folks Much much for our model hierarchies inference averaging of models is extremely powerful something I always pushes we should try to stuff the list on a mickle case law when this when this crisis is gone we should try model model as it were everything we don't know what the process is I'm up with threats are so we should really be modeling a whole gamut from the minute that we see but what we've got to not equate civilians to be able to do that either it's other ways it's really tearing code need to just finish off I talked about Paul in my poly microbial interactions and these are really interesting a dramatic example of these that are a lot of dengue a mix of other things that are a lot lots of poly part of microbial interactions and but as you'll probably know and this is not the work by Rachel Baker again.

[00:16:30]
Don Dunstan Daniel pocket and I as the that the typical captain of the very important childhood viral infection r.s.v. is is the dotted lines here. And. Everything really in in in in the current this is. And this is the start of the. The national emergency of the pond and I.

[00:16:58]
Don't want that and I think many other viruses have dropped through the floor but the trouble is Rachel runs this fall within the in the in the. In the paper the susceptible is going to build up right a metaphor you predict that you'll get maybe out and Noble faces some are epidemics for example received a winter disease and because you people don't get infected then they'll get infected later on so they're a long term poly microbial implications for endemic diseases of these contacts and very interesting Now this is just out in Australia they are seeing.

[00:17:36]
Out of phase southern summer epidemics which is kind of quality it's in the predicted by these models there's a ton more to do here so it's a vital and a very exciting area and the traumas disciplinary quantitative work that you folks will do is absolutely key to it Ok I'm finished thank you so much Brian Todd with an absolutely fantastic talk and we're going to hold off on my questions and we'll have a I kind of come a combined Q And a session for 20 minutes at the end of all of the talks so if people want to ask questions just pop them in the chat there and we'll be monitoring and monitoring the questions we'll collate those.

[00:18:26]
Details shall I introduce Professor you were would you like to. I could go at Ok thank you x. box so I'm delighted to introduce our next speaker of Professor been you from u.c. Berkeley. Should be talking about your reading a quick 1000 data Depository and forecasting county level that concert us so Professor us Johnstone's distinguished professor and the class of my people 6 2nd chair in the opposite just sticks and electrical engineering and computer sciences at u.c. Berkeley and a former chair of statistics at Berkeley she focuses on the practice algorithm and purity of statistical machine learning and causal inference or group is engaged in a disciplinary research that scientists from genomics neuroscience and pretty precision medicine.

[00:19:14]
She's a member of the u.s. National Academy of Sciences fellow of the American a book like Academy of Arts and Sciences and she's won a lot of awards including being a Guggenheim Fellow The 2 key memory a lecturer she was the president of the Institute of mathematical statistics and also she received the e-mail Scotto out from the committee of Presidents of statistical societies so we got further ado I welcome Professor you to talk about her.

[00:19:41]
Insights thank you. And it's like. Thank you very much for inviting me for your pleasure to be here and also learn from all the colleagues so we're kind of like an Oxford kind of jump into this coping prediction data repository curation last March like for 2 months my group and see that most of my group was jumping with me become a response to come by a nonprofit organization and in the summer of me continuing then I also got help from who not well to work this money on the merits of these.

[00:20:22]
So much mission was we're working with this is not profit organization which ran your last March response for life to really try to help waste a crisis at the time about b.p. So this isn't part of and we work as 2 people from the organization I was an Arab rage to ship p.b.s. for Temple hospital in Maine so here's an overview of what we did when we joined our guys ation there was no data so we had to kill rate and get data and use our personal networks to get on top of that just ticks and then we decide to be would at the county level predict because without as Asia needed us to help them to prioritize different hospitals so the know where to ship and we only had data and he can he live so we had such group of students just provoke I worked on different predictors transparent and simple and work on uncertainty measures using busy prediction residuals and we round up side this kind of automatic Ai system so here space of what we curated was about 7 and Southern hospitals characteristics and also at the country level we had many many together about $203.00 features for the hospitals and maybe 50 of the.

[00:21:47]
Counties including social economic factors and many other lyme ability so this is a public of people to use. I want to grab the issue hasn't come out a lot which is data quality that there are 2 sources us in fact and they supposed to be a New York Times are going to come from the seams also suspended on gray and there's also we can you fact which shouldn't be there and those of people do his circle revisions for space through our prediction I've been out of you know to a very this continues and we make more errors So this thing's out think it's the bottom of all this epidemic predictions really data quality I think we need to pay attention to so as I sat with you about a many many different predictors I took advantage some were going to the Bell Labs 20 years ago but I'll do prediction came by handy which is combine in there and we found 2 of them are used to lean a predictor and the other expanded shared accounting predict county level prediction which is use the neighboring county for a week ago the data under that and cases and then the past that's come from the same Congress over a simplistic and then we combine depending on which one works better in the past few days and we weigh them so where a simplistic and kind of engineering and signal processing to do is answer for n.t. which is also took about I can also approached that we just look at the last 5 days the prediction arrows are equal and has made and then we took the next month and then went at and plus the prediction on top of that can do some calculation and there are assumptions ability condition that we're gay 83 percent coverage if you want higher coverage you can choose to more factor in peer group because the good thing a bad thing about this is that every day you have a new data point you can kind of team the coverage and this lawsuit is called come form a prediction without probably sick generate your mom so you can see that's what.

[00:23:47]
We can do it and this is you can see this I was made almost a year ago but then it's methodology hasn't changed and then the paper was really ready to support a website and has appeared in Harvard has science review we we're not going to write a paper but we don't have because we wanted our work and because reproducible so we decided write it up and this is the website with me running the call with a very good outcome to support the response collide but also just to keep the mission now they're not a good and then shorts and a country like what the pan while we were doing it there was no other country level that sport and our predictions are Adam on the c.d.c. website so we could use our predictions to the c.d.c. Web site and if you take the 2 or click on the left you actually can get the Arlington County this is from yesterday you can see that the produce is the case number of the rand is that them and they're not not on the same scale do you have black and my skills can see the hopefully things are turning on in 10 counties in which in so to take a step back this is we constructed a automatic Ai system but every time a cubit thought I go to best site and find something's not my working right so this is really say so we need a human machine collaboration that you need a human surveillance of the bad side to correct mistakes like for example on May 21st it was and her entry of 15000 that from King County in Washington is later got the project to 513 that night so there was probably a human error but that machine with a no that and just have a huge jump and our agony just answer off I also want to acknowledge a dub you as an asset of support because we're using these 8 ws computing resources to do a big data grandpa Mansour and that's what we buy and all.

[00:25:48]
So for our team we still run a state a repository open source and it have a now mastered very simplistic transparent interval and to wear a comparable shock I'm predicting performances Asian based mechanistic models so I think we should try to leverage those approaches. Just finally got access to California hospital patient data and within some cause investigation I feel like causality is very hard to establish but we have to make a decision so in a sense it's a causal investigation because these nations even don't do anything you know change that's a decision and not a policy change so I think we really need to get evidence so how the policy maker can make an in compensation take a step back to look at the challenges opportunities we really need and in the national international survey this is the Intervention Network system and has become key to being a manufacturing supply cannot just reclining and stuck out her card nation because we're seeing people you live so.

[00:26:52]
It was not easy to work with steam a for example and with the pressure 3 which are establish a relationship with them and it is a policy it's like the most fundamental 1st step for all the other system to be reliable otherwise we might have very products cation of Arrow humorous Ishant you and other arrows that Byron how to crack up downstream and this has to be in collaboration with Ai system writing is beyond human scale so in a responsible trustworthy and reproduce we're a system with human endeavor it's by delving into the ai system a feel like we need a trans different framework as already mentioned we need a coherent a unified technical accessible terminology so the multi-discipline people can work together and we use the same terminals in thing.

[00:27:46]
And we also need a integrated distributed computing platform for that and an ethics of to support analytic like data like maybe spark and data breaks and the approaches should be diverse Asian based signal processing machine learning and with and certainty assessment also risk management capacities managed a mission reality check has to be in the loop and systematically regularly carried out and with a multiple objects health economy and finance and there's this dynamic change that nobody can predict so their surveillance and the reaction times we fast when we change things departing from what we know we have to be edgy and to adapt how can it change in other possible economic social human balance so I conclude I shall give a shout out to her remark My group has developed for the whole data science lifecycle particularly You can see what we did is a click cation of these P.C.'s framework P.C.'s means predictability computability and stability we're also working on textbook to refute and it has science on this basic principles and now a key component of P.C.'s it's actually documentations I think we also want to balance a system we need a very good documentation to document The human judgement calls as we all know the modern know when they make and for every step of the process.

[00:29:16]
So on the brand the point about us humans right is so if we don't have good human researchers all team members and we cannot do this so the culture has to change from winner take all Eugene science to collaborative culture and staff and under 20 agencies hope what have visionary and fair incentives the rewards for Grant also shape reward and just a worst and I personally think that it's not enough as I've just to support transform into research but also have to support ratification replication full trust where he and I will research and we have to really know how the human and machine should collaborate and education a speck of a trance the spirit of a search has to be why much of the Center for the young generation people to kind of get in and continue his.

[00:30:08]
Research thank you thanks a lot but call that. My stock and. Give it over to all I think or introducing the next week or if Thank you that was an absolutely fantastic talk and next up Professor Jordan Pet Care Act is going to be presenting and I suspect it is the Thomas Eagleton Jr professor of environmental engineering at Yale University his research mixes genetics with engineering to study childhood exposure to bacteria fungi and viruses can Belding's transfer packet is a member of the Connecticut Academy of Science and Engineering and associate editor for The Journal indoor air is a Ph d. in environmental engineering from the University of Colorado and is going to be talking about tracking epidemics at the population level through wastewater based epidemiology So thank you so very much Hopefully you're ready to go here I'm ready to go Grace thank you welcome thanks.

[00:31:14]
Ok again hopefully you can all see me share my screen and you can all hear me Ok I want to think you offer the kind invitation to speak today about our team's research on wastewater based epidemiology So ironically my day job is sort of to think about pathogens and buildings and.

[00:31:35]
I had some history though of thinking about human exposure to pathogens the suit sludge and when the epidemic started you know there were 2 different pass to go and I had to pick one in and we went down the path of focusing more on wastewater based epidemiology So here's an experiment that we did several years ago and it answers a seemingly obvious question but a pretty difficult one to get you that is what type of microorganisms would keep a viral pathogen specifically existing wastewater it's difficult to know because those are difficult to culture especially with all the different strains we didn't have the genetic tools capability and he was billeted in databases Intel just recently but when we were able to put all that together deep sequencing databases and genomes for different microorganisms we were able to get a bigger list of what each of viruses exist in wastewater this is interesting because you can think of cities of a 1000000 people or more all being contributed into one single wastewater treatment plant that flows into one area and one thing that was interesting to note was the high concentration of coronaviruses indicating or suggesting least that they were got tropic organisms.

[00:32:46]
And that led into you know answering some questions on initial cult phone calls in March of 2020 about whether this could be detected in wastewater and many groups went through the same sort of evolution and there are a lot of groups now working on this idea but the general idea is is that you can take effluent concentrations waste water as it goes into a treatment plant or some other area that leaves a building in our case we look at sewage sludge you can determine the concentrations the source code need to emit by p.c.r. similar to the type the test is done for human testing and when we did that early in March we found that there was a peak that corresponded.

[00:33:26]
To the beginning of the outbreak and as things come down that peak went down as well these are just various measurements over here on the left in the red in these are measurements of new cases if we formalize it a little bit more you can see there's a concurrence between the viral r.n.a. a new reported case and applying some more rigorous physical analysis you can see that early on in the pandemic we were able to detect these viruses earlier then cases were being enumerated about a 7 day leading indicator due to the slow nature detecting cases.

[00:34:01]
Not only can we as recordings between this data and new cases but there's also. Agreement between the epidemic outbreaks in curves and hospitalizations at least in shape with different types of offset hospital station occurs later as well this is enough information to put together what we're calling the alcove in 1000 wastewater project this was the need through the c.d.c. and through the Connecticut Department of Health with lots of collaborators the School of Public Health and a good agricultural station and under the certain premise I want to introduce you to is it wastewater tracks cases but is more efficient and accurate than testing now I'm certainly not saying that we shouldn't do testing and of course understanding that an individual has an infection is very important for isolation etc treatment but we are all short of data and I think that in terms of population data wastewater can make some strong contributions 1st it's inexpensive right now we're monitoring more than a 1000000 people we can have a cat taking 6 samples per day each of these samples cost about as much as it takes cost to do an individual testing it can be fast the thing is still in many cases prompted by symptoms and it takes a while to test that person to get that data to schedule testing to get the data back and report that we can take in wastewater data on Monday and we can report that data to the public later that same day and in some cases in some ways it can be more accurate certainly than testing surveillance doesn't care whether the symptoms Aisam that whether it's a semantic or not is long as they're shouting but it also doesn't care about things that really impact testing and testing volume like holidays governments snow sources snow storms and all the things that we've been dealing with there is a relationship between these 2 and if you can get to a testing program that is robust in a wastewater testing program that is robust You can compare the 2 and this is just through a simple regression.

[00:36:01]
That mostly depends upon late concentrations of r.n.a. in the waste water and some adjustments for the strange testing behavior on the weekends. You know we get coverages that are above 9.9 in all cases across different wastewater treatment plant for the same model and some of these treatment plants are order of magnitude differences in sizes what is going to meet him immediately and we I guess have to think beyond this pandemic or think beyond this country is that if we didn't have ways of estimating cases for example for New Haven Connecticut we could estimate them throwaways water in the curve would look like this the curve actually looks like this this is the actual cases this is our estimate of cases just from wastewater data same thing in Norwich Connecticut this is what we estimate the curve to be simply from wastewater data this is what the real curve looks like it still captures the same type of behavior it also.

[00:36:58]
Is smoothed a bit more because you know sometimes we regress in else's or pigs mothers so another practical way that we can do this in the framework that we're working with now in cases and what we have is it can keep our state up to date our state uses keep by cases by data specimen collections we don't use things that our New York Times or our.

[00:37:20]
Or even that are reported are state we look back at the date of specimen collection and that means that again there is delays in that taste test data so let's example say take Bridgeport on February 15th we had wastewater data for that day and all the days previously we were able to estimate the cases in that's was represented here in these light bars we get all the way to the black bars which is about a week earlier this is the most updated case by date of specimen collection so it's very useful for us to tell the state.

[00:37:51]
We know that you don't know what happened you know this entire week but we can have to meet in roughly tell you yeah cases are still going on this week. Who have about 3 minutes left. So. We look this it into policy we report this. Work to the commissioner of health and state epidemiology we talk 2 times a week we work with c.d.c. we work with health directors of hospitals and we have a front facing website to the public you can get there by e-mail could waste water it looks like this you can click on the towns you can get the Arnie concentrations you can get the keys by Doesn't these collection there's all kinds of different ways of looking at it there are protocols for people to use as well a few other things that we can do with wastewater data and this is with Sarah needs and at the Connecticut egg station if we can look at markers of behavior beyond infection so this is just driving with traffic looks like bends or try is all as a hitter this put in brake discs to inhibit corrosion that leaks when brakes get us down to the roads rain washes it off it gets into wastewater treatment plants and as indicated we can see when the state or state orders stopped or started.

[00:39:02]
The traffic going down when we opening went up traffic when I see the same thing with opioids that in many people have looked at this type of data before opioid use increased with the outbreak except the opioids are used for elective surgeries we have early warning systems now where we're able to look retrospect we stuck to flee back and determine that based on when outbreaks occurred and what the wastewater concentrations looked like and how you would model wastewater concentrations in cases together we can take a threshold these thresholds of yellow are 5 cases 400000 people and when he says we're low over the summer and we hope cases are low again over this next summer we're able to provide 90 percent confidence to mayors to health directors to say yeah you know you've just crossed a threshold of 5 cases for 100 we're not 90 percent confidence that something but we know this when we did cause these crises thresholds before we went from 5 cases 200210 cases prendre 1000 to 15 cases 400000 then eventually to have more than 100 cases perhaps.

[00:40:07]
The last thing here I'll tell you about is genetics and. There's a real opportunity here for surveillance of genetics we have a severe lack of capacity to test do genetic testing right now for specimens that are collected in from looking at variance in without through and through the state that I live in but also without you know throughout the United States and I just want to show a comparison here some specimen testing early on in the pandemic separating the New York late in the Washington d.c. clade here's the same thing at the same dates on wastewater and showing that we could pull genomes out of wastewater we could separate different types of clade but there's a lot of problems here this is coverage from this is coverage in the coverage is pretty weak we solve that problem by just focusing now on the spike proteins and I think moving forward we have some good solutions for doing that so the last thing I'll leave you with is if I think that we need to move forward the 1st thing we need to do is is move toward automation and there is no inherent barrier to that surveillance of other infectious diseases these are things that don't have testing programs for sure is where it can be useful or in places where our source code to do testing is not great.

[00:41:21]
We contract in extremes and evolutions on a more population level through wastewater we can even do things like look at gene expression in human cells as worker disease and modeling all the modeling you think about doing with cases in the problems with cases I think could also be done with wastewater data that doesn't have some of the same issues and finally you know we don't have time to talk about this but you know don't get me started on what we could do in the developing world and how important this could be so here's all the participants it's been a big collaboration if you want to find the wastewater Web site you can go here you can put that you are and you can google it or you can go to my website in click at the feet of my Doc thanks.

[00:42:07]
Thanks thanks a lot. Out of all that really great Dr Paul I guess we can go ahead with questions yeah and so this is actually going to be a combined q. and a session for everybody that was speaking speaking this morning and so kind of a mix up of a q. and a session and on the panel.

[00:42:28]
So we've had a couple questions come in through through that chance and Krista one of the questions is from you actually if you want to ask us. So one initial question was if major restrictions will cause a larger breakout occulted $1000.00 later how be predict the magnitude and time of the next breakout wave and will such restrictions cause more or less total infection anybody on the panel that would like to take that one.

[00:43:01]
How can we predict the magnitude in time of the next breakout wave and will restrictions cause more or less total infections I mean it depends how the immune dynamics work we're only just learning that. The thing that I showed was about another infection at the end against. Covert mounting causing to drop down I mean I would have thought that was doable but it's very interesting wonderful talk by my colleagues it's it's wonderful to see Jordan start because again.

[00:43:37]
Using wastewater might be a really interesting way to get a handle on Actually Brian while you're while you're at talking there's a question that might be most appropriate for you which is that recovery from measles meant permanent immunity and why is that not the case for covered. So it's going to be health with permanent immunity and there's almost no such thing as for many is infections really sterilizing meaning t c You see a slight bump in the antibodies if you're exposed to keep it if you've had measles and slightly mysterious it misses it's incredibly invariant the Surface Pro teams where the weather where the weather to get that it has to get into 2 cell types it's very invariant it's a system systemic infection problem isn't one reason about but it just seems to be transmissible but evolution really stuck which is great news because if there was loss of immunity measles.

[00:44:31]
We would have considerable problems I think the vaccine isn't quite as good and there's always a people keeping an eye on what box immunity waning but we're not really seeing it very strongly yet but it's a great question if actually that I was actually listening to our Science Friday when I was in the car and Friday and a listener had asked Dr Fouty Why can't we have vaccinate against h r b as possible I think it is got people thinking about well what's the next thing we can vaccinate against which is which is really quite confusing Can I ask Can I ask can I still open a question let's talk about a lot of talking and such and such a great system.

[00:45:14]
Have you looking at other viruses or other pathogens and it would be a really interesting time now to look at pathogens which has really been knocked down by the Neponset Ic an intervention against Cove it and then see if you can get an early warning about the way that they bounce back particularly out of season I think that would be a really important thing to do in a very elegant study.

[00:45:37]
I think that that's all a great idea the you know one limitation that we would have in wastewater is that not all of the pathogens that we care about have have got trojans and more enough control Bisan to show up in waste water. I don't know the reasons and I haven't thought deeply about it but we don't seem to see a lot of influenza in wastewater.

[00:46:01]
And we see as we see a lot more coronavirus then right of my mistress Molly and my suspicion is that that has more to do with cell tropism than it does. Nice sure you know. If we can figure that out I think we're going to be in a better footing to extend to different viruses and Terra viruses might be interesting but not have it as serious So you're saying **** there's a little bit of that's right yeah thanks Rhonda and John but a nice discussion so I had one question about and this is for all speakers their comments do you have to you all talked about video surveillance programs right so I remember back talk about you know surveillance at crosscut surveillance John obviously from the risk water and also talk about radios ways of getting case comes so hard how much do you have to David concept of this is that to operate in a sense of machine learning which is in this a real estate up program so like to give you an example there have been a syndrome experience programs of weapons up to grants to try to deal with the better the relationship of the surveillance data to the disease indicators suddenly change try to novel events maybe a little bit of a people are using.

[00:47:17]
Very little truth and scary side maybe here I don't know if there's a festival or something or there's a huge influx of people coming to the talk so hahaha you have to be the how much you have to deal with these kinds of sudden obvious changes in the syndrome except when dictators thanks you know I would say I would say that that's a really great question.

[00:47:43]
And I don't think that I can get quite maybe to the answer that you're in there that I think you're looking for in our case but but partly that's because because our experience is and we just have tried very hard not to let the concept drift. Because we just you know we think that the thing that we started with is fundamentally important and we just don't have the bandwidth to chase all the other really interesting.

[00:48:09]
Ideas going on so that's that's part of it but I would say the more just you know thinking about the data that you get in surveillance. You know we're an academic lab like everybody else here is I think when we find ourselves monitoring right more like a commercial lab and things like supply lines in availability of reagents and how that changes in how trends and doing analysis changes and equipment changes are real and they really impact that and it's it's very hard to push all those continually back at the science evolves and say that we're still monitoring the same way so I think that's where we've had a lot of difficulty.

[00:48:56]
And I see. Are you going to. Our army make anyone else want to join I've been wanting speakers Yeah Our think that for me to waste just the situation can change that dynamic very fast or close the concept just the situation happy was behavior the only way we can get ahead it's really very agile like surveillance and has billeted to act by pay 60 monitoring and regular quality check on the data and with human in the loop right you have trained a human to do these checks that just like if you look at traditional manufacture something and it's also creates this and that right so they have this quality control.

[00:49:39]
Procedures in place probably simpler than what we do expect the same philosophy shop life here is it a regular checking by humans and discussions of communities I think a jump certain questions are in the chat box about how do we arrive at call here in unified access what I mean ology what I had in mind was like in her composition manis and they actually have this category own dictionary of medical terms some come media Queda and there's different from medical claims you have this cast vacations will become easier we have to get together just have a document and be reviewed periodically also Baxter's and that's the way I think people can refer to and go for young people come in that's the way I think we can go make a.

[00:50:31]
Difference. It's a community and a city find some dictionary terms. Thanks for responding to that question I wonder if I could ask a further question Dr Peck Yeah very fascinating talk. I'm guessing that the r. and a that you get is extracted from virus particles because I'm imagining it would degrade if they were just naked aren't a waste water.

[00:50:58]
Does anyone look at the the. Infectious activity of the nerves florist particles at all because viral is just commonly are aware of only a small fraction of total particles being exception so there's something called the particle to show does that anyone ever make this kind of measurement for the particles from wastewater.

[00:51:25]
For SARS Kovi to that that's been done pretty rarely and even if you go back into the more obvious case and say just looking at. Infectivity is inst in human still you know for a long time the dogma wise there's lots of my wrists are in a way it may be viral particles but nothing is in fact if I'm not sure in fact as I'm not sure that that's true it's awfully difficult to test for infectivity and still in wastewater Yeah.

[00:51:57]
It's my understanding and I think Christopher knows this feel pretty closely as well she may know that there are some cases where viruses when there's infections terms can be 2 in human stool but nobody's found it in wastewater and nobody particularly is looking because it's I don't think it's a high value target for the people who have the capability of doing this right thanks so.

[00:52:27]
Bondy do you want to ask a question yeah I did actually because it's an essay behavioral economics sciences question so it's actually from Brian but I think Professor Ross or you might be able to also had chime in as I saw some some G.I.'s now visualizations on your You're really amazing that sport and so the question is how do we integrate social and behavioral sciences into these models at institutional level had we guess Quantitative Social scientist we gauge social dynamics behavior change etc fundamentally change transmission dynamics and complicate control this seems like it should be a priority but we don't see a lot of progress and maybe if you could just talk about what what you need from the social behavioral sciences what what could those disciplines be providing in model form or in quantitative signatures that that would help you do more with your work so let me tried so we did have collect some you know public social economic factors right and then we put the arm although it just didn't help so I think has to be a most sophisticated way of understanding the dynamic wells of putting some ability into and you seem to be Ok out in bad it into the neighbor counties case and that's number that's what gave us the best predictive result and that's only for prediction but if you want to go for interpretation then there is policy change so that's what we have some project you have a little bit not completely cause experiment going on there because the counties make different decisions and train their policies so I think that's I think we have to formulate a team maybe we should talk afterwards is how do we establish sound causality like evidence that replace a lot of this neighborhood called the cops that's what we use interest something more interval.

[00:54:24]
We are learning what these do to help the prediction so we wanted a simpler model but if you do this tradition is this is good testing many of these demographic and social economic factors what turned out to be in a significant which I don't exactly trust so we just decide about our prediction for the purpose they worked we support the p.v. disabusing we're not mechanistic So I think we have to integrate some of these mothers in law spatially come here many are beige and Iraqi women know when I was just in the way my signal processing way.

[00:55:02]
While country at a time because we had a kind constraint of I had to do things in a few weeks just just expanding out what's going on something my great colleague at Princeton Simon Levin is very interested in this and we have a great student. The young is looking at the dynamics of mask wearing and it's a massive dater about different countries and different trajectories of must must wearing and one of the determine the standards of the social norms of government actions and so on I get the impression the pandemic like with many things it's a really sweet time to think about this because a lot of search for scientists.

[00:55:49]
Really great economist resample are interested in our knowledge of those sorts of data I mean I have a question for them I would it would be how does a particular intervention for example model a transmission people risk aversive their risk seeking and even qualitative answers like that would be kind of useful I think it's an exciting time.

[00:56:12]
I think there's many ways you could do it as well one way would just to be to build an agent based model of absolutely everything that would have a one to one mapping to reality and possibly we're possibly getting there you know in the future maybe in some areas.

[00:56:25]
I know your ex your ex colleague at Princeton Yanis can record this had done a lot of really great work on equation free modeling So the idea that you can pick up signals of some of these other other systems and project them to a manifold and then from that manifold you can get into it into other models and that's that's quite promising had to work with him and some some civil mechanical engineers and trying to do that for Asian base model and there's lots of ways that we could possibly do it we've gotten 32 participants on this workshop here so I think together we can probably kick start kick start something that could help some promising research along these avenues I think when you bring up it's really how do we integrate the age in baseball those with the mission learning thing a process and approach I mean is the time we had to couldn't really get into that a teacher to know when you have to work with a group already have an agent based model when I sell a size that we have we can do very well thought of the trends to do I shall term prediction the company kind of use that to constrain the aging based models beyond just choosing the you know the social network and all the other parameters you too so use it well as a concentrating that.

[00:57:39]
To come by that way I knew I wanted to point out. So I got to be far I've spent quite a bit of time and research thinking about microbes in buildings and not always very specifically about viruses and things like influenza or common cold or rhinovirus But you know for many years everybody has known the tremendous human health toll from respiratory infectious disease and and you know part of my tradition comes from environmental engineering where you know we would never tolerate that type of disease level in drinking water but we do tolerate that in the indoor environment for some reason or another.

[00:58:23]
I am hopeful that you know one of the silver linings that could come out of this and Demick is that we no longer tolerate that in more and in order to do that we're going to have to break some of the rules that that have you know and a way of doing things that we've done for so many years in people the natural sciences are just going to have to find a way to pull in people from social sciences or allow them to come in or whatever the barrier has been you know we're never going to get there with respiratory disease unless we have the contribution from the social sciences and behavioral science and I just thought about I mean I agree you have to be really perfect that right because the susceptible is a kind of build up.

[00:59:07]
And if it gets rejected strawman animal species you're in trouble again in some sense so I agree with you but lots of accidents I missing a much because well you know we've got multiple ways of controlling. And that this has been an excellent panel and I want to just tell you a moment just the knowledge once again and to find all of our speakers for the remarks that you provided and for helping to inform our conversation Spartan web years and new ways of thinking and also for taking the time to be responsive to all of the questions that came in from our women our participants or thank you all very much again we're hoping that all of it will be able to continue to remain with us a little while longer to participate in the breakout sessions.

[00:59:57]
We know that we plot to continue. Benefit from your perspective thank you once again.