 Awesome, thanks so much. So in the context of this presentation, my goal is relatively to provide a short overview, a brief overview of some of the geospatial methods that are potentially applicable towards understanding some of the questions that we've been talking about in the context of this workshop in terms of understanding neuroscientific processes or just processes of interest in situ. And what I hope to accomplish in the context of this presentation is the relatively modest goal of illustrating some of the tools that we have at hand at the moment to understand how some of these processes of interest operate outside of the lab, outside of the clinic in the context of everyday life. So building off of some of Mike's presentation and understanding how context enables us to answer questions that relate not only to between individual variation, such as, for instance, how it is that Vietnamese Americans, say Los Angeles, differ from Vietnamese in Hanoi, but also within person processes and variation, such as how it is that one person differs from context A to context B from time point one to time point two, and if there is no difference or the variation is low, then what that means, which I personally find to be a little bit more interesting. So what these methods get at is this general notion of putting people into place. And what this essentially gets at is this observation that physical and mental health vulnerabilities, as well as outcomes, are not randomly distributed but often geographically clustered. So let me provide a few short examples. So one of the most commonly studied place effects has to do with living in impoverished neighborhood. So in terms of the general literature, what we know is that residents in more economically impoverished areas, on average, have higher all cause mortality than do those in wealthier neighborhoods. And so, for instance, traveling from the wealthiest neighborhoods in London, say, to the most impoverished area, you see that male lifespan expectancy drops by more than 20 years. So, and concomitantly, individuals in impoverished neighborhoods are estimated to spend approximately 17 years of their life living with a disability of some kind. And so these gradients emphasize the needs to understand how it is, the context sees in which we work, the context sees in which we play, the context sees in which we live, influence health outcomes. And so part of what putting people into place really means is how we begin to understand these different exposures that people have on a day-to-day basis. And what are the methods that allow us to look at these exposures with greater granularity? So place effects. Really, I think there's been increasing interest on place effects, but really, as in meta-analysis in 2012, with Suggest, most of the studies that examine place effects really just focus on residential neighborhoods. But anyway, what it impacts, place can impact a range of outcomes that have been documented in literature, such as allostatic load, self-rated health, sleep quality, symptoms of depression, anxiety, somatic symptoms such as dizziness or nausea, as well as a range of other health outcomes. And also that place structures, network opportunities, and thereby potentially risk-taking behavior. So as an example, among injection drug users, for instance, 94% of their romantic partners and 97% of their drug use partners reside within a five-mile radius of where they live. And so I think this really draws on some of the conversations we had on Tuesday with Lawrence and Carol in terms of the company that we keep and how it is that place both enable and constrain these interactions, who we associate with. So in terms of what space or place really means, I think really it is a little bit of a black box that is similar to the way in which culture is a black box. And indeed, we could potentially think of place effects as encompassing three different types, three different types. And some do in terms of the compositional effects that have to do with the attributes of the individual that reside within that space. So for instance, the religious affiliation, the age, and what have you. And then there's the contextual effects that have to do with physical as well as the social disorder of the particular space that we're looking at. So for instance, attributes like broken windows or the number of syringes or the number of venues that are available for sex work or for alcohol consumption, say. And then there's also the shared meanings and practices and attitudes within that particular space. And so I think at least for the sake of this presentation, perhaps we can briefly think of space and place as being this special ensemble of physical infrastructure, technology, information, resources, and social relations that is physically grounded in a place on earth. So that is there are coordinates that one could look at and encompasses processes that unfold between people as well as between people and places and between different places as well and the meanings that people attach to their location. So for instance, particular cities, for instance, is listening feelings of particular areas of a city, listening feelings of fear, feelings of unbelonging or alienation, and what that really means, residing in those particular locations on physical and mental health. So I think when we're thinking about place effects, this gets us back at socio-ecological models of which there are various kinds, but for the sake of memory, for memory's sake or for, I don't know, emphasis, memory, reinforcement, I thought I would bring back Carol's bio-ecocultural model, which focuses specifically on the developmental niche and how it is that we begin to understand child's developmental outcomes in the context of the developmental niche. So for instance, the child's own capacities and capabilities vis-a-vis caregiver psychology and also in terms of caregiving practices that are pervasive at the time, but this of course is nested within a larger cultural and physical context in which of varying house conditions, of varying exposures to violence, of varying exposures to infectious microbes that can influence a number of later health outcomes. So, and I think these, what these socio-ecocultural models, whichever one you look at really aims to accomplish is really to extend beyond these proximal individual determinants of health towards, to situate them within particular structural, organizational community, as well as familial contextings and to really understand the dynamic interplay between the individual and their environment. So I'll be speaking a little bit more about that in the context of experience sampling and how it is that you use experience sampling in the context of more geographically explicit approaches, but also more, I just thought I would go over a different socio-ecological model, Rhodes's socio-ecological model that has a little to do more with what I do, which is in this case HIV. So on one level, you know, you have the individual with their individual attributes. So for instance, for the sake of my research, say an 18-year-old, men who have sex with men, men who have sex with men, rural urban migrants, Hanoi, whose Vietnamese, male sex worker, and this individual might have very particular perception of their risk environment, of their space, of their relation to Hanoi. For instance, a lake, Tingguang Lake, which is this very popular site for sex work. And during the day time, you have male sex workers, female sex workers, sorry, that surround the lake. And if you were, for instance, an Asian female researcher anthropologist, say, that were to do a field audit of this site, then you could easily be mistaken as a female sex worker. But during the night time, this changes completely. During the night time, say 10 p.m., for instance, this suddenly shifts to a site for male sex workers. And so you have men, old men, that are cruising the scene in search of male sex workers. And often also, you know, and so there's the risk environment changes. And this person's relation to their perceived risk environments will be different, for instance, than a 60-year-old female tourist from the U.S., even though both might be at the very same site, because Tingguang Lake is actually aside from a very popular place for sex work. It's also a very popular place for tourism. Not sexual tourism. So just to clarify. But anyway, so there's the perceived risk environment, right? So these are all functioning at the individual level. And the individual might feel, for instance, socially isolated by perhaps a more concrete sense, a physical danger. And indeed, you know, among the sample that I work with, 30% or so, have had a history of sexual abuse. So as individuals become hyper-villageant, attention becomes biased towards social threats and also negative expectations independent of actual risk exposure within, say, Tingguang Lake. And this perceived sense of physical and social disorder have been shown in the literature to corrode, to increase hyper-villageants, as well as to corrode perceptions of neighborhood quality, which can predict worse health outcomes, which can predict increased cardiometabolic risk and higher allostatic load. And as I've mentioned, with the example of the 94% of sexual partners and injection drug users, 97% of drug partners residing within a five-mile radius of each other, there's also a second level to this, which is, so there's the individual level, but then there's also the social networks that people are enmeshed in, which might promote different risk-taking behaviors, such as alcohol consumption, such as smoking, which can increase, for instance, inflammation. And in broader strokes, these are ways of thinking about nestedness, and I think also the applications of multi-level models towards understanding this nestedness of individuals within different contexts. So we have implications for intervention. But these models, of course, are very fancy and they look beautiful. And of course, obviously, the more that you put in there, the more beautiful it becomes. But really, when you're measuring it, how do you simplify it? And what do, I guess, geospatial data, what do GPS-related data sort of look like? And so, I mean, maps really simplify reality for analysis, and therein is the cartographic paradox. And to quote, it really, to present this beautiful and accurate map, what you need to do, a good map, needs to tell white lies. So, for instance, in, that image is not visible. Okay, there we go. So, for instance, in my left, that image right there, you see that a house may be thought of, it may be portrayed in many different ways, but here in the context of this map, it is a red dot on here. And this facilitates our ability to calculate distance. So distance from the home towards a clinic, distance from the home towards a school, or distance from home towards a food store, for instance, and a road and river, it might differentially divide, but here in the context of this map, it really just constitutes one line that divides two places, that form the boundary between two places. So, I think what this really gets at is that, however complex the models, however complex we think about nestedness, geospatial data really boils down to two things. One is the spatial data that specifies the location where someone physically is on earth, and the attribute data, which have to do with the features of that particular space. So for instance, this building is located 45.5 degrees north and 75.6 degrees west, for instance. But attributes of this, for instance, the width of the building material, as well as the number of windows, say constitutes attributes that are stored in a database table. And so the research question should drive decisions on levels of aggregation, scale, and type of data that you use. And this actually presents several dilemmas in terms of decision making in the context of research. And two of the issues that usually come up in the context of GIS related work has to do with one is the modifiable area unit problem, which is this relationship between context and outcome. It may differ depending on the geographic scale or the zoning area chosen. So for instance, let's say that you're interested in the effects of green space on everyday mood. But the question is how should you define a neighborhood in this case for analysis? And that is intricately tied to this idea of the modifiable aerial audit problem, a unit problem, sorry. So do you use census tract to divide these places into neighborhoods? So for instance, do you use census tracts or post codes which don't necessarily map onto everyday experiences, but nonetheless, because there are census data, they allow you to also add on survey data that are readily accessible. Or on the other hand, do you want to define neighborhood by drawing on, for instance, local histories or how the individuals residing in these particular spaces would themselves define neighborhoods? And so findings regarding this are a little bit mechs, actually. So there are a number of studies that show that actually how you define neighborhood actually doesn't seem to matter from the outcome. But on the other hand, you have perhaps one of the most widely cited study that I see relating to this problem, which is by open trial in Taylor in 1979, which showed basically that depending on how you just sliced up counties, sorry, how you sliced up neighbor counties, yes. Then the relationship between Republican voting and the number of old people in that district could, the correlation between which could vary anywhere from negative 0.97 to positive 0.99. So that's a big difference, right? It's opposite sides of the spectrum altogether. So these are real problems that really ultimately boil to what it is that you're interested in and how you should begin to conceptualize context. What really is context? What is the environment by which you will derive your exposure measures? And the second problem is the uncertain geographic context problem, which actually taps into what I said earlier in terms of really, oftentimes, even though we speak about context, we're actually not really that clear on what we mean by context. That is, how exactly is context affecting? How do places exert effect? And sometimes we don't often know what is operating, for instance, in that neighborhood that generates these outcomes that we're interested in. So this is also tied intricately to the fact that we don't necessarily know where people are, how long they've been there. So for instance, we might know, for instance, that Bob resides on the street in Atlanta, but we don't necessarily know how long he's been there and we don't know where he is. We might well not, well, maybe we do know where he is, but we don't know how long he's been there. We don't know where he goes about on his day-to-day basis. So for instance, so actually adults, among adults, 70% of their time is spent in non-residential neighborhoods. So on a day-to-day basis, Bob, as he goes through work, as it's daily activities, for instance, might actually be traversing through a lot of places with different green spaces. So taking it at the neighborhood level is not that easy. And then also, multi-level models are great when there is a clean nestedness. That is, when, for instance, families are nested in, individuals are nested in families, families are nested in particular neighborhoods, neighborhoods are nested in particular districts, but at the same time, the hierarchical linear modeling, multi-level modeling, isn't that useful when there isn't that clean nestedness, right? So for instance, on a day-to-day basis, you might go out to work, which is outside of your neighborhood. You might go to shop in a particular space, which is outside of your neighborhood. And also the same individuals, because they have different activity spaces within the same household, may have different exposures. So this is one illustration of this uncertain geographical problem that we have. And so this is some data that deals with substance use behavior. So say what you're interested in at the moment, what you're all interested in at the moment, is substance use behaviors. And you want to predict substance use behaviors by people's everyday exposures. In particular, you thought it would be great if you could think about exposure in terms of an index for community SES exposure. So this is basically a drawing from census data where you're trying to create an index of things that has to do with community SES. So for instance, with median income, for instance, household income, or the number of households with parents with heads of households with college education, say. And then the second thing that you're interested in is violence. So on a day-to-day basis, as your participants are going about their daily lives, what is the degree of their exposure to say low community SES to a lot of violence? And how does this predict their substance use behaviors on a daily basis? So say in the context of this study that you decide to recruit a number of individuals, 47 of them, for the sake of illustration, and these individuals, you give them a GPS logger. And so this GPS logger, they'll carry around with them for the duration of the study. And this GPS logger will log a coordinate for every 20 meters that a participant goes, or alternatively, every 15 minutes, whichever happens first. And you also give them a palm pilot, or you just have them install an app on their phone. And these individuals are going to answer questions three times a day at random points of the day that ask them about drug cravings, that ask them about substance use behaviors, that ask them about their moods, levels of stress, and so forth. And also you're going to basically also ping people, you're also wanting to make this event contingent. So you're going to have people report when they're using substances outside of the three random time points. So let's say that you track these individuals for an average of 107 days. So after about 107 days, you're going to have about 35 million GPS data points. So what do you do with the data points? How do you measure, like what here, how will you measure exposure, right? So you have this GPS, these 35 million GPS coordinates that tell you the spaces that people traverse. So what is the context here by which you'll derive your exposure, being the community SES? How much, for instance, during those 107 days, they were exposed to what level of community SES to violence. So there's different ways of slicing this, right? One way of doing it is that you could look at the GPS coordinates themselves. That is to say, for instance, if the first coordinate is in a space where community SES is at, I don't know, five. And then on a second GPS point where it's at a 10, then you're going to average it. And so that's going to be 7.5, and that's gonna be what you say is the exposure. The context here is going to be the GPS coordinate, and the exposure is 7.5, based on the context that you've arrived at or decided to use. Alternatively, you could also say, for instance, you could imagine that maybe these GPS coordinates aren't that accurate. So for instance, in highly dense neighborhoods, in highly dense cities, in ideal conditions, iPhones, the GPS coordinates are about accurate to about nine meters or so. So you don't trust it that much. So you decide to create a buffer zone around each GPS point. So that's one way also of measuring context, and that could be the context. Alternatively, you could draw a polygon around that would encompass all of your GPS points. And this could be your context, and from there you could measure exposure. Well, how you define that context has a lot of implications here, as you see. Now, here, as you see, for the variation that you see in people's exposure, in this case, to community SES. So imagine that this, the higher this is, the lower the community SES is, and the negative here is corresponding to higher community SES. And so these are different approaches that you could take of, well, I guess, putting context onto this GPS data. So one way, as I've mentioned, is to actually use the GPS coordinates themselves, which is the blue line right here. But alternatively, HB, right here, is to use the inferred home. So basically, to use the inferred home address. And from there, to use that as context. And this will differ, so for instance, in this case, if you're using people's home as context to determine their exposure to community SES, then you see a lot of variation among these 45 individuals that you're interested in. But alternatively, if you were to draw just this polygon that encompasses all of this space, so this is the red line right here, you can see that there doesn't seem to be a lot of variation actually. And so to put some numbers onto it, in terms of if you're doing inferred home location, that's the HB, this black or dark blue line here, then the center deviation among your participants is about 6.5. Alternatively, if you're going to use inferred home address, and that's the red line right here, much flatter, as you can see, across participants, then the center deviation is about 1.55. And the same could be said in terms of the impact of what you choose as context on crime exposure could also be illustrated here, where some approaches suggest actually there's not a lot of variation between individuals in terms of their exposure, whereas other approaches suggest that there's a ton of differences between individuals on their day-to-day exposure to crime. So this matters a lot in terms of, for instance, the variation that you think of that exists in exposure, and also a lot of differences in terms of the conclusion that you arrive at, whether, for instance, location or residence actually impacts health outcomes. Actually, in the context of this case, so it might influence it, but it might not. But it's to suggest that these things matter. And it's a real problem to consider when you're thinking about what context is and how you should study context. So this is, unfortunately, a slightly outdated meta-analysis that showed the rise of place effects publications. And so you can see that there's been an increase of that and no doubt still an increase, although 90% of which, as I've mentioned, really just focus on residential neighborhoods, and most of which focus on the effects of poverty and living in impoverished neighborhoods in particular on health outcomes. But nonetheless, these studies are increasing. So it does raise up the question of why now? Why be interested in place now? Why pick up this now? What's that? Why bother, right? So I think part of it has to do with this idea that locational precision is very strongly tied to statistical precision. And with the proliferation, as Mike mentioned, of smartphones, you have now the greater availability of low-cost GPS that's relatively accurate, as I've mentioned, maybe nine meters off or so. But so for such spatial methods, what you require are locational data that's accompanied by some kind of attribute data. And now smartphones are increasingly available. So in 2017, about 80% of Canadians own a smartphone and about 70% of Americans own a smartphone. And not only do they own a smartphone, they're very invested in their smartphone. They carry their smartphones around a day-to-day basis. So for instance, approximately 82% of Americans report that they don't leave their home without carrying their smartphone with them. And not only do they carry their smartphone with them, they're actively engaged with their smartphone. So the average amount of time that people spend daily on their smartphone is about 2.7 hours. So this suggests an opportunity. And this suggests potentially the use of smartphones as a scientific instrument that could potentially also deliver interventions. Who really knows? I mean, I think we talked a little bit about some of the ethical issues regarding technology, regarding, for instance, and no doubt also the veil of smartphones in particular. But I think one thing to consider is, for instance, in low and middle-income countries. For every 3,000, for every 5,300 individuals in low and middle-income countries, there are approximately 2,300 smartphones, but only 11 hospital beds. So it does perhaps suggest the potential for digital health interventions, the potential for smartphones, in perhaps negotiating with some of these stark dynamics. So now I'll transition a little bit to talk about the different types of geospatial data that are available. So one is more at the census and population level. So this is an example of AIDSView. And AIDSView is an online interactive map that's made available by collaborators at Emory. So Patrick Sullivan and the Prism Initiative at Emory. And it allows for users to explore the HIV epidemic in the US by state, by county, by zip code even. And so why is this important? Or why be interested in this? So data from the CDC, for instance, show us that in 2014 that of those who are living with HIV, 86% know of their diagnosis. They know that they're positive. And 40% are engaged in care. So that is they've gone to at least two clinical visits. And 37% are currently using antiretroviral therapy. But only 30% are virally suppressed. And time to diagnosis can be estimated based on the disease progression, the CD4 cell count, at the time of diagnosis. And as you can see on the top graph, there's actually a lot of racial disparities in terms of the time to diagnosis. So for instance, Asian-Americans and Native Hawaiians are much more likely to be diagnosed later, as well as older Americans. And they're also actually more likely to, for instance, Black and Hispanic men who have sexist men are more likely to have current STIs. And this will increase the risk of HIV infection. But these dynamics are not evenly found within the United States. They're regional, as well as state-level differences. And 8th View allows for us to be able to visualize some of this. So for instance, in this graph here, what you see is the percentage of males living with HIV that's attributable to male-to-male contact. And the ones in dark are where male-to-male contact is the primary mode of transmission, whereas in those in lighter counties, so lighter shades of red, other modes of transmission, such as heterosexual transmission or injection drug use, are more common. And this changes, for instance, when you consider male-to-male sexual contact alongside injection drug use. And there's also a means of visualizing the HIV care continuum, which is also powered by 8th View that allows us to see these disparities that exist even within Atlanta in terms of HIV diagnosis, in terms of linkage to care, engagement in care. So what this shows right here is engagement in care. So individuals that are going to clinics and after diagnosis. The areas that are darker suggest lower linkage to care, whereas those are lighter suggest higher linkage to care. So perhaps to no one's surprise, in the city of Atlanta, for instance, linkage to care is much better than it is in rural areas that surround Atlanta. But actually, it's when these things don't quite match up, that it's also interesting. So for instance, in Vietnam, where I work, there are multiple outpatient clinics where an individual, a diagnosis of HIV, could go to get their HIV medication. However, preferentially, people go to OPC and TL, which might be in some case. So currently, there are 199 MSM that are receiving treatment at this OPC. And the distance that they travel varies quite a bit. So you might be able to justify why it is that someone might go five to seven kilometers to this OPC, as opposed to the OPC that's one to kilometers from them. But what about individuals who are going 100 kilometers, 150 or 200 kilometers, which I see? So what's driving that? And so I think on the one hand, we could be thinking about this map as allowing us to see issues of accessibility. And where it is that we need to leverage resources, where we need to leverage interventions. But accessibility can be conceptualized in many different ways. On the one hand, it's accessibility in terms of whether it's actually physically there or not. But there's also accessibility and whether it's within physical traveling distance, and also whether it has the services of interest. But there's also issues of accessibility that relate to acceptability. That is, whether the services themselves are acceptable, whether they're perceived to be friendly or not. And I think when there is this incongruity between access, different forms of accessibility, it's when it's particularly interesting. And which GIS can help us illuminate some of these dynamics and ask questions about why that might be. Another study that I think also operates at a population and census level that's also rather interesting in terms of its use of GIS-related data is in thinking about the relational aspects of space, of place, that is not just about the physical attributes of resources that are there or not, but also about how people relate to one another and the historical contingency of space has to do with this study that looks at cardiovascular disease in the United States. And this is out of some of Michael Kramer's work in Atlanta. And so at Emory, sorry. So in the United States, cardiovascular disease continues to be the primary cause of death. However, at the same time, there has been a decrease in heart disease mortality by 62% within the United States. But this decrease in mortality is unevenly distributed. So for instance, in certain southern counties, such as in Atlanta, for instance, as compared to somewhere more up north, such as in Boston, what you see is that in southern counties, the reduction is about 50%. Whereas in more northern areas, you see that the reduction in heart disease mortality could approach somewhere like 82%. And also on average, the decline for white Americans in terms of heart disease mortality is much steeper than it is for black Americans. And so why might that be? And so there's a lot of hypotheses that one could leverage. But what Michael Kramer did was to look at the legacy of slavery, which I think is sort of interesting in terms of how he used GIS-related census-level data. So what he wanted to test was this hypothesis that it had to do with the country's differential history or engagement with slavery. And so the legacy of slavery differs in the United States as you know. So in some counties, slave ownership might be 0%, whereas in others, it might approach 95%. On average in the United States, it's at 29% or so. And so what is of interest is that perhaps these historic institutions, beliefs and practices, emerging from differential slave ownership might influence previous or ongoing disparities in terms of location of resources, of technology, of information that might be driving differential well-being and thus contributing to material disposition or distress among this population that's forstalling this decline in cardiovascular disease mortality. So what they found is illustrated in these two maps right here. So what you have is Moran's eye, and this has to deal with spatial autocorrelation. And so you can interpret this in the same way that you can interpret, for instance, a correlation coefficient very roughly, which is that it ranges from negative one to one. And the idea being that the higher it is or the higher the absolute value is, the more that things that are close together, clusters together. So there's a clustering of outcomes. So here, this one looks at heart disease, disease decline among black Americans and this one among white Americans. And you can see, and this right here has to do with the history of slavery within the United States in slave ownership, whether that concentration of slaves approach, for instance, in lighter colors, just zero to 7.1% or whether it exceeded, for instance, approximated 50% in the darkest shades right here. And you could see, for instance, that these two maps look vaguely similar to each other. And indeed, that was what Michael found, which suggests actually that slave ownership was positively associated with population size, land value, and manufacturing output in 1960, higher black illiteracy in 1930, as well as black poverty in 1970, and concomitantly it was also associated with a slower decline in heart disease mortality among this population. And I think that's rather interesting because actually due to this ongoing flow of black Americans outside of the South, you can't actually say that it's really a, you can't say definitively that that's because these individuals or children or descendants of individuals who were enslaved. And this is why we're seeing this outcome, that it has to do with some sort of intergenerational trauma, but perhaps it actually points to more of the longstanding effects of this historical period and how it's influenced some of the institutions or some of the practices, some of the regulations that operate at these county levels today. And what he actually shows is that the sort of secondary outcomes such as black illiteracy and poverty actually, in 1930 and 1970, actually explained 50% of the variation that you see in heart disease decline among this population. So really interesting stuff, and this just goes with what I've said in terms of slave concentration being negatively associated with heart disease decline among black Americans within the United States. So let's say that after what I've said, you become interested in GIS related stuff. So how would you find these sorts of data? And so I think there's many sources of finding readily available data sets, GIS data sets. There's the secondary data that are readily available and that you can get online as well as the administrative records, for instance, from a clinic or a hospital, as well as primary data that you can collect. And as you go down this list, the specificity of the data to your research question increases, but also, so too, to the cost and the time that goes into it. So going from the census level, population level, now I'll talk a little bit more about neighborhoods and how people have sort of worked with neighborhoods and neighborhood effects and how does the one defines boundaries as experience and perceived differentially by people? So on the one hand are the administrative units, the census tracts, the postal codes and divisions by counties for statistical reasons. So as I mentioned, this is useful because it allows you to draw from readily available census and survey track data. However, what you end up assuming is that people living in the same neighborhood have similar exposures, even though one person might have lived there for an X number of years and one person might have recently moved there or alternatively, one person lies in the periphery of that neighborhood whereas one person is at the center, but there is one neighborhood effect. Nonetheless, this is where the majority of the research really is when we're talking about place effects. Alternatively, you could look at presumed activity space. So there's been some work, for instance, that tries to look at, that looks at the effects of children's exposure to children's exposure to disorder, physical and social disorder on antisocial behavior. And in this case, to define the presumed activity space of the child, they use accelerometer data to show that actually on a daily basis, children operate around 0.5 mile radius of their home. In which case, if you have that kind of data, then you could define that activity space as being this 0.5 mile radius around your participants' home. Outsortively, another approach is to look at perceived boundaries. That is how the residents themselves perceive or experience these types of neighborhoods. So example of which is Colton et al in 2001 and what they basically had was to present individuals with maps of their home at the very center and then eight miles on all directions and then have individuals sort of shade in where they considered their neighborhood. And what you see is among individuals that are living in the same census tract, the same block, sorry. There's about a shared shading of about 70% of the area. So this brings up interesting questions as to the differences in terms of what's perceived as common and what's different and what that means about those particular spaces or how people experience them. And this is, I think, a really good example of neighborhood effects. And this actually comes out of Candace Audre's work and so actually quite closely relating to what Mike was talking about. So one of the most, as I've mentioned, investigated effects of place effects has to do with poverty and specifically living in impoverished neighborhoods and what this does to outcomes like risk-taking behavior like antisocial behaviors. Because, for instance, hypotheses, sorry, theories, such as broken window theory or opportunity theory would suggest that physical and social disorder would embolden individuals to engage in these kinds of behaviors in contexts in which they're normative. So policies encouraging, for instance, max neighborhoods draw from these types of hypotheses to suggest that, perhaps, proximity to high-quality resources will benefit, for instance, disadvantaged children. But of course, the alternative hypothesis could also be true in terms of relative deprivation and relatives SES, for instance, where daily reminders of one's lower social positioning might actually contribute to worse health outcomes. And so Audre's, in this study, is trying to disentangle those two effects by investigating the influence of neighborhood SES on U.S.'s engagement in antisocial behavior. And this drew from a longitudinal study of children in the UK over a course of 12 years. And they measured antisocial behavior at age five, seven, 10, and 12. And again, this draws this in terms of how they define context. It was also, from what I mentioned, in terms of the accelerometer data, which is showing that daily activities for children around 8 to 12 or so is within a 0.5-mile radius. And so this is how you get this circle that surrounds the individual's participants' home. So the red dot is the person's home themselves. And then what Audre's and colleagues did was essentially to categorize, based on census data, neighborhoods or these blocks into either those that were wealthy and well-off, so characterized by high income and single-family homes towards those that are more hard-pressed. So for instance, where the head of the household had no educational qualifications or, alternatively, was receiving government benefits, aside from disability benefits, where annual income for the household was less than 10,000, for example. And so you can see two examples here of two different participants. One is subject one, and so this is their home. And you can see that this is a hard-pressed person. It's a hard-pressed household, but they're surrounded by healthy achievers, wealthy achievers, or alternatively, those who are at least comfortably well-off or relatively well-off. Whereas, for instance, in subject two, you have this individual that's being surrounded by other individuals that are hard-pressed. So what does this do for antisocial behavior? And so this is what I suggest, what the data suggests, which is right here, for instance, when you're a disadvantaged youth that are surrounded by wealthy peers, and you see actually an increase in antisocial behavior. But whereas, if you're a hard... But those individuals who are themselves hard-pressed but surrounded by individuals who are hard-pressed, then antisocial behavior is actually lowest. And this is actually the opposite effect. It's observed for their non-disadvantaged peers where being surrounded by hard-pressed individuals actually increases antisocial behaviors. And I think many ways of thinking about these findings, but it certainly highlights, for one, is the importance of sociality to place. Because on the one hand, there is the physical deprivation that has to do, for instance, of being surrounded by less resources, by, for instance, houses of worse conditions, even worse conditions. But there's also the psychosocial dimension of poverty that derives from these interactions with peers that surround oneself, and which imparts a more personal side to SES that might escape statistical capture in this case. So I think this raises a fundamental question as to the processes through which people become emplaced and the methods that are available to get us at the more proximal questions of how it is that places effect. I think these are very pertinent questions because, as I've mentioned, one of the most investigated phenomenon regarding place effects has to do with poverty. So Dan Linde, for instance, has an article about this notion of how poverty poisons the brain, and it takes a critical stance on that sort of literature. And so I think these are important points to raise. So as one might imagine, well, I'll skip over that for now. But what if the data are not available? So when we talked about, for instance, the population and census tract data, those data are readily available. In terms of looking at neighborhood effects, for instance, in Autres's study, those data are also readily available. But what about when, for instance, the data that you want to answer the questions that you want are not readily available? What do you do? So in which case, well, you have to go out and collect it yourself. And so there are different approaches to this. One is a field audit. And field audits are basically rating this respondent-centered area. So you know where your participants are, and if you're interested in that activity space around them, that context around them, then you could go yourself, and using a checklist of predetermined features, looking at social disorder, looking at physical disorder, social decay, social activities, begin to map out what exactly is the surrounding environment for your participants. And you would need, well, a paper or smartphone, some sort of GPS log or some kind. And this often requires two or more people, so you can get at inter-rater reliability, and then also to do multiple audits both during the day and during the night. So, for instance, some things you'll only be able to observe during the nighttime, whereas others, you know, the same thing would hold during the day and the night, for instance, like such as the number of abandoned buildings versus something like social activities. Like what are the adults in that particular space currently engaged in? This might well shift depending on the hours of the day. In terms of the tools, the questionnaires that are available, so these are just examples of which. One is Nifty, for instance, and this is what has been used in the Baltimore study to look at physical and social disorder and how that influences, how stress influences substance use behaviors. There's also the active neighborhood checklist, which focuses more on activity levels and the neighborhood attributes inventory as well as other inventories. But in terms of the tools that are available to facilitate this data collection, this audit of the field, if you will, there's a range of apps that you could use on your phone, for instance, so like the Open Data Kit, which is the most, I think, commonly used or what is, I guess, the oldest one, the originator for everything else that follows it, which is a free open source software that enables for survey development and survey data collection, both online and off, so that's particularly useful when you're working in resource constrained settings in which internet might not be reliable and there's a large community of users and developers, but unfortunately there's a very little customization that would be available and it also only works on Android. You could also use the Kubu Toolbox, which also is derived from the ODK and this is an online builder that allows for people to answer questions very easily on the phone and on the web and what you have is an app that you could download and it works on iPhone and Android and it allows for data collection offline, so basically people download the survey when they're online and then they complete the survey multiple times offline and when that person has Wi-Fi access again, the surveys are finally pushed up into the server and so this is also a great option to use, when for instance your participants don't use Android that much and prefer iPhones and you also need the offline option. CompCare is by Demagi, which is increasingly, I think, used in the context of digital health and they're great in terms of it's free but with paid options. So Demagi does a lot more customization options that are available for you and it allows for tracking and what is particularly interesting in this case that is perhaps not so much relevant towards doing field audits but are more relevant towards using geographically explicit EMA is that it allows you to track specific individuals over time in real time. So for instance you can follow a case, for instance how Bob feels on day one, day two, day three in real time and be able to see this on a very user friendly interface. Unfortunately it only works on Android but also quite great and then RedCap which I think if you collect health data you're very familiar with, most people are more familiar with the web option but there's also actually an app option. To what degree that's actually a very stable app that's a different question I think but certainly it's available and then next is Biwi and so some of my work in Vietnam is using Biwi and so this is more of a digital phenotyping app so this is less about field audits but it's more about if you want to phenotype individuals, if you want to identify behavioral signatures and how it is that people are using their phone and being able to see how that associates with a range of mental health outcomes for instance then Biwi is more for you, it's more patient centered, it's not really for field audits but I thought I would just mention it briefly. So I just wanted to provide just a very brief tutorial on what this would look like which may or may not be very useful so let me know if you just want me to skip over. But Elmo is also one of these apps that could be used for field audits and so it's maintained by the Carter Center and this is a platform that I've worked with the Carter Center to adapt for use as a tool for field audits as well as for participant driven EMA and but originally Elmo slash Nemo was developed by the Carter Center for the sake of election monitoring as well as human rights violation and it still maintains a lot of this in terms of how it works so it builds off of the ODK actually if you actually use the mobile version of this as opposed to the web or tablet version then it does run on ODK and it simplifies the survey process and adds means of monitoring data in real time which is great when you need to see these sorts of data in real time and this is what the user interface this is sort of what the backend looks like so this is particularly why I mentioned that it might be useful for field audits because all of this is readily visible to the researcher and it also is geo-coordinated and it also allows for you to easily add on researchers or participants who are helping with the field audit so quite useful and convenient for those who find it so and next is the mobile app which again as I've mentioned runs on ODK and this is sort of what it looks like one is that you have the administrative you have questionnaires that you could build in so these are forced choices that people make and then also in terms of recording your GPS location so this is where people just press record and it records the latitude the longitude of where this individual is and it also gives you a sense of what the accuracy is so as I've mentioned in ideal conditions in urban cities the accuracy is about within nine to 10 meters or so and this changes depending on where you are and these are things that you need to consider when you're using these types of apps exactly how accurate are these low-cost GPS sensors that are readily available on smartphones so this was a so as I mentioned field audits in the context of how I've done them is that I've employed staff who are health educators who are MSM themselves to engage in field audits and certainly you could also have participants engaged in field audits and this is one study of the community alliance for research engagement that has individuals living in those neighborhoods actually do more of the auditing work and here just identifying food stores, restaurants and the like so in terms of Vietnam so this is an example from Vietnam as the title would suggest and so this is looking more just this was just when we were interested in doing a field audit of popular sites for sex work male sex work in Vietnam and so these are locations where we do outreach actually so daily outreach where we go to these sites and we're able to see and count actually number of male sex workers that are there and so that's the dots that are yellow or green versus the ones that are darker color and then we were interested in doing a field audit around these sites of sex work so what this looks like in effect is well one is that you would select some sort of field audit questionnaire and so here I used the Nifty questionnaire and this is a 169 item questionnaire that tries to get at physical and social disorder and so youth activity and adult activity during the nighttime and daytime and also indicators of alcohol and drug use and so the presentation yesterday the very last presentation said that the world is saturated with pilot studies and this is very true especially in so too is the life of a graduate student because in the context of this presentation I think what this example highlights is actually not only the application of field audits but actually in some cases where they're not really that useful or great because they're so expensive and actually very time consuming so here in this case when we were using Nifty we were trying to map out the area that surrounded for instance a massage parlor where sex was sold and so when you're doing these sorts of field audits at least in this case, you're walking here so the first time we're going so say imagine this to be the starting point we're going up this way once just walking normally to see what's the distance that we can cover in 10 minutes and then we walk back to note the layout, the structure so for instance what are the types of residences that are available, the non-residential buildings the residential buildings, the businesses that are in that particular location and then you're doing another walk for another 10 minutes where you're noting physical and social disorders so broken windows for instance or the number of fights that you witness the sort of activities that people seem to be engaged in so what are the adults doing or the adults men doing are there adult men that are loitering around how many adult men are loitering around all kinds of these sorts of questions and then you're going back a second time to monitor again physical and social disorder because again there are so many indicators that you really need to double check whether your account is accurate and then you're going a third time this time to note down the youth and adult activities and these sorts of audits well you need to have two individuals that do them at the same time and independent of each other and then you also are going to do it well not just once but twice and then also during the night during the day to see what are the changes from day to day from day to night and in effect this is sort of what it looks like when you're doing these sorts of field audits so this is what it looks like in Vietnam and this is the old quarter area and this is an example of the path that you would sort of take when you're doing a field audit around a massage parlor and actually so this looks like a very small space that's being covered indeed it is a very small space that is being covered but it took us about four hours to do one of these and so four hours twice and then twice day and night so that's about 16 hours total for one location and of course this is only a very small segment of where individuals are going on their day or are on their day to day life so it's very difficult to actually measure in this case activity level and to have any concrete or robust measure of exposure to physical and social disorder and of course we're not the only ones who have difficulties with this so there's been more work that looks at field that goes beyond field audits to look at well can we use what's available to us virtually can we use Google Street View for instance and so this is some great work by Candice Audres so she's also has what she calls a neighborhood dashboard that actually aggregates and is a tool that aggregates online information related to neighborhoods so it's not only just what's available on Google Street View but also from census and then also from patient from participant reports and from a range of other sources to be able to build the most comprehensive possible sense of what neighborhood exposures mean and so an example of the application of these virtual audits is again with the environmental risk longitudinal twin study in which 94% of neighborhoods for participants in this study are viewable online and so you can train raters living or working in the UK to go around virtually and where participants are and to be able to again using this activity space of a 0.5 mile radius to be able to categorize or rate disorder, decay, danger and street safety so you know disorder for instance in terms of graffiti that's visible in terms of abandoned or burned out cars decay in terms of physical decay again these are very physical indicators dealing with street conditions for instance and deteriorated gardens and what have you and then danger just having to do with unsafe places to live street at night so basically asking the raters themselves to rate how safe they would feel walking down this area at night and then street quality so again these are very physical these have to do with physical decay physical disorder because you're using because you can't monitor social activities when you're doing a visual audit virtually but what they found actually was that these indicators that were assessed virtually mapped on to the census data so this is the acorn so this is the census data were significantly associated with census level data and then it was also associated with local resident surveys so local resident surveys are basically mailing surveys to residents that live around participants' neighborhoods living alongside participants and asking them to rate what is the neighborhood problems and neighborhood dangerousness and it seems that actually these virtual indicators are also mapping on to what residents on a day to day basis are experiencing so I think that's interesting on two fronts one is it speaks definitely to perhaps the utility of these visual online audits but it also suggests that actually local perceptions perceive neighborhoods rather than just being very abstract or heavily biased actually do seem to map on to some sort of physical concrete reality of some kind and so in terms of inter-rater reliability it was 15 of the 17 items exceeded 70% inter-rater reliability but the only two items in which this was not the case is sidewalk quality and also safety to walk at night so more subjective measures and I think this is also true of other online visual audits that have been conducted in the past which shows that actually these more subjective senses of safety and danger they don't, the agreement around them is relatively low so I think it really depends again on what it is that you're interested in but I think for the sake of this study at least it shows that actually these virtual audits are these indicators that when arrives at using virtual audits actually do map on to child antisocial behavior pro-social behavior and BMI so interesting stuff but there are limitations to using Google Street Views which is that actually this works fine when Google Street View is readily available so in the context of that study 94% of participants' neighborhoods are readily available but this isn't always the case this, you know, Google Street View is not evenly available within a country it's certainly not available globally so what you see on this top figure map right here is that these areas that are in dark blue are those where there's complete availability of Google Street Views and light blue are those where there's partial availability and green are those where popular tourist spots or destinations or main streets are available you have photospheres that are available and then in gray or white are the areas where none of this is available so, you know, this works well when it works well below is actually an example from Vietnam so you see these dots right here it's a little, the resolution isn't too great unfortunately but here's a dot, here's a dot, here's a dot, here's a dot and these are photospheres where Google Street View is available so you can see these major streets where it's available you can also see right here this area surrounding the lake where it's totally not available and this is Hanoi, this is a major city in the capital of Vietnam when you get to a smaller urban city like Nha Yang for instance which is still in the north which is still a major city in the north you actually have only one photosphere from the entire city and it overlooks this one shop and that's it and you have nothing else that's available so you can't really do visual audits in this case and I think, you know, I think there are interesting questions to be raised and should be raised regarding the politics of representation in terms of what's available and what isn't available what's covered and what isn't covered and so one really good example of this is actually an article by Powerdoll in 2012 which look at the availability of Google Street View in an estate in Ireland and this is an estate that is characterized by 12 parks and over 1000 households and when Google Street View was just available in this area in 2010 the only places from which it was available was always from other estates that overlook this particular area and the views were always of abandoned streets of high walls that separated the estate from the surrounding areas because this estate was characterized by high crime and violence and so it suggested a very particular view of this place so in any case, that's also one I think issue to consider and the second issue to consider is the degrees to which individuals are actually able to accurately identify their residential address on Google Street View so on average among individuals that have some familiarity with maps it's about 0.65 miles and this is a little bit worse in Vietnam whereas people might be relying more on certain queues like where the old cathedral might be where Jangdeung Plaza might be and so in which case there is a larger margin for error so again, so what we've been talking about is census level and then neighborhood level data that you either have to collect yourself or that's available but another thing to consider is activity space so activity space has to do with the spaces that individuals cover on a day-to-day basis and why this might be important is that for instance, Americans on average travel about 29 miles and also where people live don't necessarily coincide with where they are where they play, where they work, where they go and so forth as I've mentioned about 70% of times that American adults, 70% of the time in their day-to-day life they're outside the residential neighborhoods and this is also higher among, for instance, adolescents where 50% of those who are 15 to 29 actually spend 92% of their times outside of the residential neighborhoods and for children ages two to eight most of the time they're spent in the school days sorry, in the weekdays in schools whereas during the weekdays they're at entertainment or food places that are also outside of their neighborhoods so when we're interested in activity space what we get at is spatial polygamy which has a great definition by Matthew and Yang here and spatial polygamy is this simultaneous belonging or exposure to multiple nested and non-nested social and geographic, real, virtual and fictional and past and present contexts and the idea really is simply that people move and exposure over a person's lifetime differs and while neighborhoods have generally been the focus of substantial place-based effects there's no correct scale at which to assess place effects and I think increasingly we're moving on beyond neighborhoods as being these discrete units towards seeing it as a more, as continuous data so a really good example, a really cool example of this is actually out of the Baltimore study that looks at substance use that looks at physical and social disorder and how it maps or actually doesn't map onto drug cravings and drug use so this is a study in which you have participants who rate on their smartphones on a day-to-day basis three times a day their moods, their levels of stress, their drug cravings and these are individuals that are currently receiving methadone treatment and also you ask them about levels of stress and the physical and social disorder is actually done through field audits so this was an extensive three year long audits of physical and social disorder in neighborhoods in Baltimore and what's interesting about this study is that it shows that if you look here drug heroin craving as well as cocaine craving and negative and positive mood what's interesting about this study is that well, you know, based on theories like opportunity theory and window and broken windows theory what you would imagine is that physical and social disorder would corrode neighborhood cohesion exacerbate feelings of social isolation, fear of physical danger and this would create perhaps higher negative affect higher levels of stress and these feelings of higher levels of stress and negative affect might potentially lead to higher drug cravings for cocaine for heroin for instance but what they found interestingly was the opposite which was that in areas characterized by the raters the field auditors by higher physical and social disorder people actually seemed to be happier in places where there were greater disorder and also their drug cravings were much lower so interesting data here so you can look at this right here so right here you see for instance this association between cocaine craving and social disorder and actually as social disorder increases it seems that cocaine craving actually decrease and so too as physical disorder increases both cocaine craving and heroin craving seems to go down and negative mood actually also seems to go down so why might that be and so I think shifting more now towards how we get at perceived aspects of space is one way of thinking about it so this is an example I think that I tend to use quite a bit and this is a study in Jerusalem looking at religious minorities and how religious minorities in Jerusalem relates to the city itself and so this is basically a map that's presented to participants, women in this study and so they're asked to draw out places in which they felt at home or places where they felt comfortable or not completely comfortable, very uncomfortable or even fearful or places in which they're not very familiar and then you give these participants a GPS logger to go about their daily lives and to see so these are the black lines right here to see the spaces in which they traverse and what you see actually among minority groups at least the spaces that they have to traverse and the amount of time that they have to spend when they're working, going about grocery shopping or what have you in places where they feel not completely comfortable or very uncomfortable as much higher than for the majority religious group and so what this suggests is that the city may evoke a multitude of feelings for different individuals and constituting varied mental maps or everyday geographies that may or may not constrain what they actually do or how they feel or engage in risk or how it influences risk-taking behaviors so if the ratio of disorder to outcome is not one to one then how can we begin to dissect these sort of boundaries and segments these flows of everyday life and then these sorts of the narratives, the feelings the stakes that are invested into space or places and you know the cognitive approaches especially out of in anthropology as well as out of feminist geography or relatively common well not relatively they are quite used quite a bit more so than say in public health example of which is here friends is provided here and here is a study in which you're asking youths in Los Angeles to map out spaces again in which they feel comfortable which is marked in green so these spaces right here right here, right here, right here and then the ones where they feel uncomfortable or even scared and which are marked in red versus the ones where they feel relatively neutral about or they're not very familiar with at all and that's marked in green right here and in the context of this study they were after the mapping activity you were the participants used Google Street View to be able to locate these spaces of belonging or unbelonging these nodes of safety or what have you and talk in a more focused group session about how it is, why it is that they marked the spaces that they did and I think this enables us to better have a more enriched understanding of safety and danger signals if you will so I think this is another example that has to do with crime mapping so having individuals living in these particular neighborhoods map out for themselves places that where they perceive there to be problems in terms of crime level activities and I mean oftentimes aside from the geospatial data that it provides it also provides really rich qualitative approaches and I'll be talking a little bit just very briefly and then very broad strokes about some of the stuff that we're doing currently in Vietnam so some of my work in Vietnam as I hinted to is really look it's focused on HIV and specifically on HIV and PTSD comorbidity and what's interesting about that is that in terms of HIV comorbidity is comorbidity is actually higher among those who are HIV positive so it's higher than what you see among former war veterans what you see among cancer survivors and what you see among the general population so in the US PTSD prevalence among those who are HIV positive is at 30% and this is again higher than what you see among the other groups and so this is remarkable and while the neurobiology of PTSD in the context of HIV is relatively poorly understood there's emerging understanding that would suggest that there's multiple points of interactions and first one of the hallmarks is intrusive involuntary recollection of traumatic memories often associated with hyper arousal and exaggerated fear response and so this is what we're trying to get at here to understand how it is that HIV modifies PTSD symptomology so this is data from Vietnam, preliminary data from Vietnam comparing those controlling for trauma what you see is that controlling for trauma exposure among those who are HIV positive compared to those who are HIV negative avoidance symptoms are significantly higher but you also see elevation but not significant of a re-experiencing and hyper arousal symptoms among HIV positive individuals and then also in terms of the psychophys side of it several psychophys indicators have robustly predicted fear symptoms among trauma exposed individuals including heart rate, skin conductance and heart rate variability where you see for instance elevation of skin conductance and heart rate among those who are exposed to high levels of trauma but also decrease heart rate variability and dysregulated startle response and this pattern is actually mirrors what you see in terms of autonomic activity for after HIV infection and so this is some of the psychophys data so this is measuring, this is very light and hard to see but basically this is measuring skin conductance during baseline and then during a trauma interview task and this is looking at their skin conductance response so you can see this right here is during baseline they're at rest relatively low and when they're talking about their trauma response it's very high and from some preliminary EM day data that we have we also see that among MSM in comparison to their heterosexual peers you see in terms of their daily experiences when we're asking people about six to eight times a day for about a two week period they're much more likely to report feelings of anger, sadness or anxiety and much more likely to report higher levels of stress and so what we became really interested in is thinking about that cognitive map and this idea that perhaps disorder isn't map, physical and social disorder isn't mapping on to perceived senses of safety and danger in a one to one way so really thinking about how it is that we could use these cognitive maps and translate them outside into the context of data to life and actually integrate it with geospatial data in more concrete ways and one way that we're doing so is to think about these zones as being zones of safety versus danger for instance so for instance out of some of the work from the Grady Trauma project would suggest that one of the most robust indicators with respect to PTSD is the failure to inhibit psychophysiological arousal in the context of safety signals so could we translate that into the context of day to day life where we're using these cognitive mapping activities to determine places of physical danger, places of safety and then sending people out into the wild so to speak when having them people wear these bio patches and so these are the patches that we use they're the bio stamps and these the bio stamps monitor skin conduct sorry not skin conductance heart rate, heart rate variability as well as respiration and with the end points, the latest version of it you're also able to monitor physical activity levels and sleep quality and seeing how this becomes altered for instance when people are in zones of safety versus zones of danger and how for instance the response to self-reported stressors varies and so we also, so we recently adapted we recently developed, adapted a BIWI for Vietnam and so this is again as I've mentioned a digital phenotyping app so we bio-engineering student that was working with me and helped set up this app and what we're interested in is again determining these behavioral signatures for instance that deal with phone usage that deal with motor activity and whether we're able to identify how that becomes modified in the context of trauma or PTSD or in the context of HIV for instance and whether we're able to distinguish individuals controlling for trauma exposure those individuals who are HIV positive versus those HIV negative based on their reactions to stress in a real time setting so data collection for this is ongoing so for the main study we have about 122, 120 individuals we're trying to get to I'm trying to get to 200 people so not really presenting anything much there and my laptop is dying so I will go to the last point which is that geospatial mapping can be used to analyze both the qualitative and quantitative research questions about place that are exploratory, descriptive, explanatory and predictive they enable us to understand how it is that individuals become in place and sort of the processes at work by which places affect people and in bringing up important questions regarding context and specifically what it is that we mean by context we were trying to translate these neuroscientific processes in situ or understand them in situ so with that any comments or suggestions or comments or questions? I think it's a good starting place if that's a real good simplification of how people actually spend their time and all these really sophisticated techniques for more real-time spatial locations is really good to see I was thinking though in the 8th it seems like your assumption is they're often not in their neighborhood they're in some other physical place and what I couldn't help thinking was actually in the age of the smartphone we're often not in any physical place and the smartphone technology and some of these communications technologies have the ability to take us out of almost any of these spaces that you're talking about so I was just thinking wouldn't it be interesting if you could combine some of these techniques with some kind of internet browsing or activity or some kind of technology to see where people are spent I mean it could be a pornographic site or it could be a game or it could be Instagram or someone in Vietnam could be reading media stories in the US but that seems like that would be quite important and some kind of metric of smartphone use combined to how people are using their smartphones is a kind of virtual place that might even further eliminate a lot of the noise that maybe you've seen some of these associations between physical space locations and help out them Brian, no, definitely I think as I mentioned with the spatial polygamy it was also really interested this notion of spatial polygamy was also really interested in the virtual spaces that people occupied but really what are the sort of methods that would enable us to do so I'm not too clear about so I think as you mentioned you know browsing history is very interesting I'm not too aware of sort of studies that track browsing activities on laptops in real time but certainly for instance in Biwi what it does among other studies in the past is with ecological momentary assessments is try to complement what you do ask with passive monitoring so things that are being observed in the background as participants are using their smartphones for instance and I think we talked a little bit about this before but in terms of for instance what Biwi monitors is for instance your call and text log it doesn't necessarily monitor like for instance you calling Bob you calling Anna or something but you know you contacting person one person two sort of thing and then also the apps that you're using and the amount of time so you're spending on those apps yeah, you're right, of course that's getting at some of the things I'm talking about yeah, that's a good stuff yeah, so actually some of the Biwi stuff it's I think they've done some work that looks at for instance in the context of bipolar disorder as well as in schizophrenia whether you're able to detect abnormal sort of usage signatures in terms of smartphone signatures and be able to associate that for instance with relapse in schizophrenia so what they found actually using Biwi was that in terms of people's phone usage as well as their motor activity so we're monitoring using these again just GPS stuff using Biwi again monitoring where people are going but actually there are anomalies in people's phone usage and motor activities or 72% higher before a relapse period for people with schizophrenia so I think these are very interesting work I think at the same time that you know the sample sizes are less than ideal oftentimes I think as Mike mentioned these are very intensive studies that sometimes don't always manage to recruit the sample sizes that we want but really good stuff that could be done with it yeah in respect to some of the disability that you mentioned physical disability oh no, physical disorder so neighborhood physical disorder well so it's very interesting because I think part of it is on the one hand so basically what the crux of this study and what it shows is that these are physical and social disorders that are rated by the researchers themselves so they're going into these neighborhoods in Baltimore and rating every neighborhood that the participant could potentially traverse them and be able to rate them on attributes of physical disorder, social disorder like the number of broken windows for instance or for instance the number of fights that you witness or cursing that you witness as you're doing these kinds of audits and that's social disorder and this again gets at nestedness of data and what you would hypothesize is that with greater physical and social disorder you would see an increase in heroin craving or cocaine craving or negative mood or levels of stress at least but they see the opposite which is in physical places that are more disordered you actually get lower heroin and negative mood and stress and I mean there's many ways of interpreting this one is that potentially that the participants themselves aren't quite attuning to the same indicators that the researchers themselves are and another way of thinking about it too is that actually perhaps it disregards context perhaps what's more important is that for instance when individuals are highly stressed they might be more engaged in heroin they might have higher heroin craving or cocaine craving irrespective of the environment that surrounds them the physical environment that surrounds them so I think that these are multiple ways of parsing through the data and I think so actually the researchers themselves so I've spoken to Kenzie Preston about this and the GIS person that works with her on this and they're also I think in some ways equally puzzled by some of what they're finding here and they're really interested I think in getting more at what I talked about in terms of the cognitive maps in terms of getting more at the qualitative these more qualitative understandings of space and what exactly people are attuning to for these individuals who are using heroin and cocaine are attuning to in their physical environment