WEBVTT Kind: captions Language: en 00:00:00.780 --> 00:00:07.680 Good afternoon, everyone. Thank you for being  here and thanks to the CIC group for organizign   00:00:08.280 --> 00:00:13.680 this and other seminars. I'm honored to be here  representing my team from Brown University.   00:00:15.420 --> 00:00:22.380 Many people from the School of Engineering, School  of Public Health, Departments of Computer Science,   00:00:22.380 --> 00:00:31.140 Philosophy, and others - too many to list on the  current slide. But if any of them are listening   00:00:31.140 --> 00:00:36.300 along today, know that I appreciate  all of their meaningful collaborations.   00:00:37.800 --> 00:00:45.240 I'm going to talk today about new - newish  projects that's just started with NSF funding.  00:00:46.260 --> 00:00:52.740 Many people - some people will be familiar  with a relatively recent NSF mechanism listed   00:00:52.740 --> 00:00:53.760 here [Predictive Intelligence for Pandemic  Prevention Phase I: Development Grants] and   00:00:53.760 --> 00:00:58.080 I'm going to speak about it a little bit, partly  because our funding comes from this mechanism,   00:00:58.080 --> 00:01:03.180 but also two of the other speakers who are with  us today will also be speaking about their project   00:01:03.180 --> 00:01:11.640 from the same mechanism. So the NIH, sorry,  the NSF, asked us to think to the future. To   00:01:11.640 --> 00:01:21.660 think in a, kind of, five to ten year span about  knowledge methodologies and data that, in that   00:01:21.660 --> 00:01:26.580 period of time, could potentially be available  that would put us on more solid footing to be   00:01:26.580 --> 00:01:33.300 able to prevent and predict future pandemics. So  that was the call for proposals - to have a grand   00:01:33.300 --> 00:01:40.620 vision of an area of intelligence that we could  address that in five to ten years would leave us   00:01:40.620 --> 00:01:48.000 on better footing to allow us to, hopefully,  prevent the next pandemic before it occurs.  00:01:49.020 --> 00:01:55.560 Our study and our project is focused on all  aspects of human mobility and social mixing.   00:01:55.560 --> 00:02:00.480 So why are we interested in mobility and  social mixing? I'll give you two or three   00:02:00.480 --> 00:02:06.720 slides on that as some background. The first is  the obvious statement that infectious diseases   00:02:06.720 --> 00:02:13.080 move from people to people Therefore, how we  move through space and time will be a major   00:02:13.080 --> 00:02:20.760 determinant on potential for a new pathogen to  spread through populations. More importantly,   00:02:20.760 --> 00:02:29.520 knowing the details about how people interact will  give us, in the future, more refined tools that'll   00:02:29.520 --> 00:02:35.640 allow us to implement more effective and more  nuanced interventions. All of us remember back to   00:02:35.640 --> 00:02:43.140 the beginning of this pandemic when, essentially,  the interventions that were implemented were   00:02:43.140 --> 00:02:50.040 those - many of them - were aimed at cutting  our social networks and limiting the number of   00:02:50.040 --> 00:02:57.540 people that we interact with in order to to slow  the spread. And because, I argue, that we didn't   00:02:57.540 --> 00:03:04.320 at that time have really good quality information  about how people interact, we were forced to apply   00:03:04.320 --> 00:03:14.280 a hammer to the situation. An unrefined tool so  that that forced, essentially, all mixing to stop.   00:03:15.420 --> 00:03:22.260 If we are successful in the future, we'll better  be able to map mixing that is conducive to spread   00:03:22.260 --> 00:03:28.440 and mixing that doesn't involve spread of new  pathogens. Therefore, we can be more nuanced in   00:03:28.440 --> 00:03:35.040 the interventions that we propose in the future. As an epidemiologist, I think of this in in   00:03:35.040 --> 00:03:39.060 another way as well. That is, I'm  sure many are very familiar with   00:03:39.060 --> 00:03:46.320 this, the epidemiological concept of the basic  reproductive number which really tells us the   00:03:46.320 --> 00:03:55.800 likelihood of a pathogenic spread. It tells us, on  average, how many new infections will occur in a   00:03:55.800 --> 00:04:01.080 population when a primary case is introduced  into that totally susceptible population.  00:04:01.080 --> 00:04:09.840 And R0, the basic reproductive number, is  dictated by three components. The first is   00:04:09.840 --> 00:04:21.540 how likely transmission is to happen between  any contact. The second is how long is the   00:04:21.540 --> 00:04:28.560 period of infectiousness. And the third is about  how frequently people who are uninfected come in   00:04:28.560 --> 00:04:34.440 contact with infected people. It's this large  piece, the C shown here, that is really the   00:04:34.440 --> 00:04:40.560 focus of our work. In essence, the other two  variables are largely biological variables   00:04:40.560 --> 00:04:46.260 that are going to differ depending on the new  pathogen that emerges. The human behavioral   00:04:46.260 --> 00:04:53.400 aspect is really contained in C here and that's,  in essence, what we focus our MAPPS project on.  00:04:54.420 --> 00:05:00.480 I'm going to talk now briefly about the main  components of our project. We have four main   00:05:00.480 --> 00:05:05.820 components, and then one overarching  sort of proof of concept exercise.  00:05:07.620 --> 00:05:16.500 First, to articulate what our grand challenge was  specifically, essentially, we're asking how we can   00:05:16.500 --> 00:05:23.820 best use data on mobility and population mixing to  inform real-time pandemic responses across a range   00:05:23.820 --> 00:05:30.300 of pathogens and under conditions of uncertainty  while still balancing benefits, risks, and harms.  00:05:31.440 --> 00:05:36.420 To do that, we focus on four different  areas shown here and I'll mention - I'll   00:05:36.420 --> 00:05:39.960 speak briefly about what we do in each  of these four different thrust areas.  00:05:40.500 --> 00:05:48.000 The first one relates to data. Clearly, there's  some, though I would argue not nearly enough,   00:05:48.000 --> 00:05:55.260 data on how we interact in different places and  in different contexts. So we're not the first   00:05:55.260 --> 00:06:02.820 people to collect or want access to this kind of  data. What we're recognizing is the existence of   00:06:02.820 --> 00:06:09.420 some data already. What we're trying to do  here is create a catalog and make publicly   00:06:09.420 --> 00:06:18.060 available online a federated database that  contains a multitude of different studies   00:06:18.060 --> 00:06:25.500 that focus on social mixing and mobility.  We hope that this will be a resource for   00:06:25.500 --> 00:06:30.720 modelers and for pandemic researchers who are  interested in incorporating social mixing and   00:06:30.720 --> 00:06:34.860 mobility into their study. [We want them] to  be able to go to one central clearinghouse   00:06:34.860 --> 00:06:42.180 and find available data hopefully already in  a format that can be used in their models.  00:06:42.900 --> 00:06:49.680 The second aspect that we focus on is around  developing devices. So we work very closely   00:06:49.680 --> 00:06:55.740 with engineers here at Brown and biomedical  engineers trying to develop new technologies   00:06:55.740 --> 00:07:02.280 for measuring mobility, social interaction, and  eventually, biometrics. Right now, we're focused   00:07:02.280 --> 00:07:08.340 on a phone app which I'm going to talk about  about in a bit, but there are potentially other   00:07:09.960 --> 00:07:17.460 devices and either other methodologies that  can help us to understand people's movement.  00:07:18.420 --> 00:07:24.420 The third aspect that we focus on is around  modeling and prediction. So the data that we   00:07:24.420 --> 00:07:33.360 collect through our apps and through our wearables  and the data that we store and process in the   00:07:33.360 --> 00:07:42.960 first thrust are are meant to feed our predictive  models. What we're hoping to do here is to develop   00:07:42.960 --> 00:07:52.800 a library of models that are flexible enough to  be able to respond to new pathogens that emerge.   00:07:52.800 --> 00:08:02.640 So we're looking at a variety of different models  that incorporate social mixing and human mobility   00:08:04.020 --> 00:08:10.440 to make them again flexible enough so that they're  not specifically COVID-focused but are able to be   00:08:10.440 --> 00:08:19.320 adapted to the epidemiological context for a new  pathogen about which we don't yet know anything.  00:08:20.040 --> 00:08:29.460 Finally, a lot of our work is around ethics.  And although it's listed as one of our thrusts,   00:08:29.460 --> 00:08:37.140 in a way, it really permeates all of the work  that we do. To collect and catalog the kind   00:08:37.140 --> 00:08:46.920 of data that we're talking about means to possess  people's private and confidential information. So   00:08:46.920 --> 00:08:54.240 we need to be exceedingly careful about how we do  that - what data we collect, how we use that data,   00:08:54.240 --> 00:09:01.080 and how we keep it in a in a safe and respectful  way. We've just finished a week-long workshop   00:09:01.080 --> 00:09:09.960 where we brought together ethicists, public health  folks, computer science people, and cryptographers   00:09:09.960 --> 00:09:18.420 to help us think about the technical challenges  that we can address when when using the kind of   00:09:18.420 --> 00:09:25.020 data that we're talking about. And we want  to do so in a respectful and ethical way.  00:09:26.040 --> 00:09:34.020 So those are the four main thrusts or areas  of work that our project is involved in. And   00:09:34.020 --> 00:09:40.200 I wanted to speak now very briefly about  an overarching proof of concept exercise   00:09:40.200 --> 00:09:45.420 that we're doing that brings together all  of these different thrusts all at once.  00:09:46.380 --> 00:09:54.120 The idea is that using the university  where myself and my collaborators are   00:09:54.120 --> 00:09:59.220 based [Brown University] that we would like  to try to measure the entire social network   00:09:59.220 --> 00:10:06.540 at the university. So imagine if everybody at the  university had downloaded our app, which is still   00:10:06.540 --> 00:10:13.020 in development. The app will use bluetooth to  measure who's in your vicinity. That is to say,   00:10:13.020 --> 00:10:18.540 it'll measure the interactions that you're  having, the duration of those interactions,   00:10:18.540 --> 00:10:24.600 and the distance of those interactions.  That allows us to build a dynamic   00:10:26.940 --> 00:10:32.400 model of the social network at the  university. What we then want to do is   00:10:32.400 --> 00:10:41.520 simulate the introduction of a new pathogen into  that into that network and assign to the pathogen   00:10:41.520 --> 00:10:46.560 some specific epidemiological characteristics,  whether that's the probability of transmission   00:10:46.560 --> 00:10:54.780 or specifically how transmission occurs,  whether that's within five feet for 10 minutes,   00:10:54.780 --> 00:11:02.340 or whatever those particular characteristics  are. It allows us to understand not only how   00:11:02.340 --> 00:11:07.500 the pathogen spreads through the network under  different epidemiological characteristics but   00:11:07.500 --> 00:11:15.540 also to identify points of intervention. That is,  things that we could change in the way that we mix   00:11:15.540 --> 00:11:22.440 and interact with others that could be effective  in eliminating or containing the virtual pathogen.  00:11:23.220 --> 00:11:29.220 Our goal here is within the next 18 months to  do a pilot within the School of Public Health.   00:11:29.220 --> 00:11:37.620 We've already been engaged in some deep community  engagement, speaking with leadership, students,   00:11:37.620 --> 00:11:46.080 faculty, staff, potential users of this device  about some of their requirements for the device   00:11:46.080 --> 00:11:52.500 and suggestions for how the device might look  and feel. We're also addressing concerns about   00:11:52.500 --> 00:12:00.720 data security and data privacy. We're hoping  to get second phase of funding for this project   00:12:00.720 --> 00:12:06.060 that would allow us to expand our work beyond  the School of Public Health and to try to map   00:12:06.060 --> 00:12:12.120 the entire social network at the university. And  beyond that, we're hoping eventually to add other   00:12:12.120 --> 00:12:16.920 measurements, including biometric measurements,  potentially other measures that we don't haven't   00:12:16.920 --> 00:12:23.040 even yet thought about, that could be used for  not only mapping and understanding social networks   00:12:23.880 --> 00:12:30.420 but also to populate models that allow us to  predict and therefore prevent future pandemics.  00:12:32.400 --> 00:12:37.920 So that's all I'm going to say at this moment.  I'm happy to answer questions and be involved   00:12:37.920 --> 00:12:44.580 in discussion about what it is that we're doing. I just do want to show this final slide here. If   00:12:44.580 --> 00:12:48.720 anyone is interested in the workshop  that I mentioned a couple of slides   00:12:48.720 --> 00:12:53.820 back where we focused on privacy and  ethics and pandemic data collection,   00:12:54.720 --> 00:13:01.200 you can find some of the talks and slides there  as well as some of the background materials and   00:13:01.200 --> 00:13:06.780 daily summaries from that workshop. So I'm  going to leave it at that and I will stop   00:13:06.780 --> 00:13:11.220 sharing and turn over to my colleagues  for the next presentation. Thank you.