 psychiatry and nursing, as well as the mental director of the road home program for veterans and their families at Rush University Medical Center. Dr. Karnick was also a staff psychiatrist at the Heartland Alliance for Human Needs and Human Rights. Narajan earned his M.D. and Ph.D. degrees at the University of Illinois, and he completed his residency as well as his fellowship in child psychiatry at Stanford. He's devoted much of his career to trying to be welcome of some of our most vulnerable populations, including homeless LGBT youth. So Narajan was a fellow here at the McLean Center and some faculty here, and during his time here in Chicago, was really one of our most beloved members here at the McLean Center. He's a very thoughtful thinker and really a sweet, kind man, and so we're delighted when he continued his association with the McLean Center as an associate. So Narajan will talk about the, we are all committed now, third parties, social networks, and privacy. Thank you. Okay, one of the nice things about the McLean Center is that despite my shifting locations here in the city of Chicago, Mark has continued to be very generous in allowing me to participate in the activities at the center, and I really do appreciate that. So I'm going to, as some of the other speakers noted, we were asked for these talk titles some time ago. And I'm going to talk a little bit about a couple issues that have been on my mind for some time because I primarily work with adolescents and young adults. And in doing so, I have somewhat unwittingly been forced into the realms of thinking about social networks, thinking about new social media and how we're going to deal with the challenges there. And increasingly, I'm thinking about interventions about this and how to leverage interventions. And this is also becoming an issue in the program that I'm running now for veterans and their families. And one of the things that I also like about the McLean Center is that it's a place that you can bring ideas to to think about and to have thoughtful colleagues help you mull about. So to begin with, I would never begin a talk at the Ethics Center without reviewing conflicts of interest. I serve as a backup study physician on a grant from Roche, and I'm a member of a not-for-profit board of directors. Neither of these positions are paid, so I will leave it to you to discern what kind of conflict I have. I won't be discussing anything in relationship to these today. I'm going to see if I can make this thing go. Before I start this video, hopefully it'll work, I hope some people in the audience remember that game, Name That Tune, right? So I'm hoping someone's going to name that social network in as few number of dots as possible. This is about a minute long. See what you think. See what you think this social network is. It is most certainly a social network. Yell it out if you know. All right, who gets the gold star? Well done. Yes, so this is a plotting over time of the McLean Center Fellows in their current locations, but the plot begins with the fellows who are the most senior in the group and continues forward. And as with any social network, what's interesting is that you see a wonderful dispersion here, Mark. Your fellows have gone everywhere. And the other thing that you notice is that the earlier fellows tended to go out further as migration is want to happen and that the more recent fellows all seem to congregate here in the Chicago area. So I thought that would be interesting to start with social networks. I told you I had a surprise remark. So what are social networks? Social networks are theoretical constructs for modeling of human relationship and interactions. They are largely driven by perspective or one standpoint. In other words, two individuals in the same network may disagree about their relationship. So we may have plotted fellows up here who don't consider themselves to be part of this family here. I think Mark is actually very good at being a good godfather type figure who keeps everybody together. So I think actually most do stay. But in the social networks out in the world, one person might say, this is my friend, Joe. And Joe might say, oh, I don't really know them that well. But in this game of social networks, things perceived as real by the individual, as the saying goes, are real in their consequences. Networks have been studied for some time now. And the origins of formal social networks research actually lie here at the University of Chicago. Ed Laumann being one of the senior people who really established the field in a really rigorous fashion. But networks are part of many disease states and pathological processes. And one of the former faculty from the University of Chicago, Nicholas Christakis, really pushed some of this work around obesity. But also increasingly, we have questions around this infectious disease and substance use. And mostly, we have studied networks from a pathology perspective, how it infers risk. But there is a flip side to that that we're beginning to explore with adolescents is how can social networks engender resiliency. And this is where I think Stacey's work actually dovetails very nicely with some of these concepts. There are many types of social networks. I'm giving you a brief primer because some of you may not have thought about this. There are three notable approaches that one can consider. One would be an ethnographic network. So our colleagues in anthropology have been studying social networks for some time. They write entire books on a single social network. They trace who is related to who, how these cultures or localities function. There are what are termed egocentric networks. And then there are whole networks or what are called sociometric networks. I'm going to show you a few examples. This would be what an egocentric network is. And so here in this diagram, you see an individual, say a patient. They're psychiatrists. I routinely engender this type of information when I'm interviewing patients in my office. Who are you close to? Who do you spend time with? Who is key in your network? I had a gentleman in my office last week who is a recovering heroin abuser and I asked him, who are you seeing right now? And he has a new girlfriend. I said, well, could your girlfriend be a source of support for you as we move forward with your treatment? Is this someone who you could do things with who could help you go to meetings? And he started to think about it. Yeah, I hadn't thought about that. And maybe that, yes, that is. But what about your parents? Oh, I don't get along with them so well. So how these networks are. And then you might ask that individual, well, what's the relationship? Does your girlfriend here know your parents or not? Are they connected at all? Do they know each other? Do they see each other? So these become very simple questions that one can pretty quickly ascertain in the clinical setting. In this instance, I've marked the red ones as higher risk individuals and this one, you know, they're more of those. Here the person has a few adults in the mix and, you know, so overall when we look at this from a qualitative perspective we might say this looks like a lower risk network. Oops, this one looks like a lower risk network here. And this one might be a higher risk network. So now you'll forgive me. I did these animations myself so I'm gonna preface that by saying here is our human subject. Here is our interviewer complete with their little iPad. We ask them, what is your social network? What, who do you believe is important to you? And they give us a vision of what that is, right? And, you know, and outline it. The reality of this is yet to be tested in some ways. I mean, we would assume in most clinical spaces people tell us generally when you ask open questions they tell you honest answers as much as possible. But we don't know the details of that. And none of these people have actually, you know in a formal way, agree to be part of research, right? These individuals are all just who you talk about, right? But all these individuals also have deniability, right? They could just say I don't know that guy at all, right? So they're out of it. So I can tell you what I think, you know about who's in my social network. I can tell you who's important in my world, right? And that can be an important basis on which to do research because of that premise I said earlier that things perceived as real by this individual are real in their consequences. Now, you can take this a step further as colleagues of mine have done and they have our research going here and what they start to do is they start to piece together different subjects, networks. So when I go into my homeless shelters and I start to ask some of the youth, you know who are the key individuals in your life? If they start to give me a little bit of information it can even be as little as first names it could be descriptors about them. I can actually start to piece together the relationships and if I start interviewing people sequentially in a setting, right? And if I ask all of you who the key people are in your lives I could start to connect your networks together here pretty quickly, right? So if you assemble enough egocentric networks so networks derived from individuals you can start piecing them together into a whole network. Now I'm going to scare you a little bit with the next step. If you get certain data, okay? For example, phone numbers, emails, Twitter and handles you can infer people who aren't known to you and who you haven't spoken to. This becomes a mathematical equation. So the human subject, the interviewer and what you start to do is you have to just hit a critical mass of information and you will start to know that there's this person in here who must be connecting these people, right? You can see the gap and therefore the gap has to be filled because the network wouldn't hold up otherwise, right? Now this gets to be more complicated and this from my perspective, you know someone who sits on the periphery of networks research community this is where they're very excited about and I have to tell you this is actually being done already by companies. Companies are already doing this type of research on all of us and we all actually have a packet number that we exist in the internet with. So something to think about is that this is happening in our world the biomedical researchers are increasingly getting interested in this. So this is an example of the whole network. My colleague Eric Rice who's at USC and I, you know, he drew on some of the data that he'd done with homeless youth in Los Angeles. So this is homeless youth. The red individuals are higher risk individuals. The white individuals in this diagram of light gray individuals are low risk individuals and the grays are sort of in between and you can plot networks, but whole networks in various ways. In this case, there's a degree of centricity around which he's trying to aggregate how risk happens. So you see these individuals all sort of hang together. They all know one another. They're interactive with one another. They may be in sexual relationships with one another. There are a variety of individual pairings that are out on the periphery here and these individuals don't have any discernible connection to individuals in the network. So this is one way that you can use network data. The interesting part for me and my colleagues is that we're starting to think about are there individuals in here that we could start to target with interventions, right? Who would then influence the rest of the network? So if I can get individuals here to reduce their substance use, right? Maybe I don't have to actually treat everyone who's substance abusing. Maybe I have to get that 10% of this network, right? That is, that are these kind of key individuals in the network, right? What some might term that they occupy nodal spots in the network, right? Or they are opinion leaders of sorts, right? And by treating them, then you propagate that change across the network, right? That they start to take that change forward without you having to treat everyone. Potentially a cost effective, potentially interesting way to go about things. Now, new social media are networks. We have to just accept that as a premise. I'm gonna quickly take you through a few examples. Most of you have cell phones. Before cell phones, you had phones at home or in your office, right? And when you did that, your phones didn't individually identify you. They may have identified your family or your workplace. Our cell phones actually now identify us almost individually. Very few of us actually share our cell phone with others. There are instances where they do. But for the most part, people individually use their own cell phones. That is something that can be traced in a network. There's a lot of research going on right now in terms of text messaging. In my area of work in adolescent psychiatry, sexting has become a big issue. I get a lot of parents who pull their kid into my office after they've been caught sexting. It's something to consider. There are also web based chat rooms, of course. This is an older model. Now we're gonna start to get a little bit scarier here as we get into the realm of Facebook, MySpace, Twitter, Snapchat is a new thing. So, I mean, the advantage of working with adolescents is that I'm constantly on the cutting edge because they come and tell me about it. And Google Plus as well. These are traditional social networks in some ways. And some of these social networks can actually plot out relationships for you. When Google suggests to you people that you might wanna connect to, it's using social networks algorithms. It's knowing who you're connected to and then it does probability scores on terms of who you might be connected to by virtue of the people that you're connected to or otherwise connected to other people. LinkedIn, Doximity are professional social networks and they'll actually show you the graphic of your network and how you're connected to one another if you punch it up there. NIH has actually started posting grants and publication data using social network algorithms as well. Finally, we get to this one. So Grinder and Blender, which I don't know if many of you know about. A lot of the youth that I work with use Grinder. Grinder is focused on the gay community and Grinder is what's called a geosocial networking system. So in this instance, what you do is you create a profile as you do with any of these social networks and you can put information about yourself in the profile. And then what it does is when the app is open, it is actually tracking who is in your social network around you, who else has Grinder on, right? And sadly, I've had individuals who use this as a way of hooking up. There's a lot of sex work going on on Grinder as well and they can quickly find each other. In Millennium Park, they can just be walking and if they have their app open, they can be messaging each other, communicating with each other. So people are concerned, I think reasonably, that these are high risk, you know, applications. But this is where social networks are going and you all are gonna be geosocialized network soon in some form. Amazon is looking to do this. Apple is looking to do this. Starbucks makes it possible when you walk past Starbucks, if you have that Starbucks app, it'll ping you, right, that you have a coupon, right? That's geosocial networking. It's tracking your movements, using your social network device of your cell phone because it identifies you. And they're even gonna get to the point where they're gonna suggest to you the drink that you like to drink today. Oh, it's getting close to Thanksgiving. You probably want that pumpkin latte and it's gonna send you a coupon for that one and get you in the door. So the federal government has given us a few guidelines on this. This is from 2002 and what it says about research in this area is that when a research project design requires the collection of information about third parties, the investigator and the IRB should consider the following factors, among others, in considering whether information is private and whether the third party is identifiable and thus by definition a human subject. So they limited the scope in some ways to identifiability. The problem with social networks is that it's kind of like the Rosetta Stone. Once you figure out one piece of the network, it becomes very easy to unlock the entire thing, right? So I'm curious, the individual who called out that it was the fellow's network, what was the giveaway in the video? Foundation and the pattern, I just recognize the pattern. Okay, so see, recognize the pattern. That's the key to this. So you can't, I would say in almost every social network, it's almost impossible to ensure complete anonymity because somebody, if somebody's within that network, right, and is able to identify themselves, they will quickly unravel the rest of the network. It's almost guaranteed you can do that. And that is the challenge in this research in some ways. How do you protect privacy? And how do you do so in a way that still allows us to go forward? And all of you have to be aware that when you sign those user agreements on all those social network platforms, you've given away your rights. You just told the company that they can do whatever they want with it. And they are increasingly interested in health data on those websites too. So my question is, how do we determine what is in the private sphere? And what is quintessentially in the public space? So I have a Twitter feed, and some of you in here and I have been tweeting about some of the talks today already. That Twitter feed is in the public square. I don't have any locks on it. There's no privacy protections on it. What I post there, if I ever, you know, mercifully I hope it never happens, ever run for political office, or if I ever like get appointed to something where you have to go before a committee and do it, I would expect all of my tweets, right, from that public Twitter feed to be fair game, all right? But I also have a locked account that I use for my personal tweets to my friends and family. That I think is private, right? I mean, I don't intend for anybody else to read that except the people in my sphere. Now, could my friends and family retweet what I tweet? Sure. And they could eject it out into the public sphere using their own means. But I think one of the things that we have to recognize is that privacy settings have to be respected in research. So if an individual does that, if you create a public account, to me that is like you're standing in the public square. Virtually that's what you've done. You're out in the open. And therein I think we have to consider the research guidelines that guide research in public spaces. We actually have for a long time conducted research in public spaces and our IRBs. And I think those same guidelines would apply there. To the extent that your behavior impacts the health of others, others can report on it in some ways, right? If I'm in a relationship with you and together we go out and consume vast quantities of cakes and tarts together, right? That is something that a researcher might justifiably wanna know. Now, maybe we can place restrictions about identifying information on that in terms of getting consent. And the interesting thing is a lot of networks researchers now are moving towards a model of using what's called respondent driven sampling, RDS as it's abbreviated, where you have one person who comes into the research study and that person helps you recruit the other people, right? They go out and help you bring them in and then you consent the other people. So it's actually using the social network as a means of recruitment. It's a very interesting model that I think allows for consent and appropriate consenting without sort of gathering too much data, but it also makes use of the network model. Finally, I think we also need to study the way privacy expectations have changed in the wake of all these technological innovations. Because I think there's a real generational difference. The young people I work with have a very different concept of privacy than what their parents or grandparents did. It's generationally different. And I think that's something we actually have to consider as we move things forward. There are a couple of people and entities I want to thank. None of these people are responsible for the viewpoints I've expressed here. University of Chicago IRB, on which I sat while I was here, a faculty was a place where we often debated some of these issues in relation to networks. The McLean Center Fellows I went through training with were wonderful. John Schneider, who's a researcher here on HIV and I have had numerous conversations about this. Laney and I have had a number of conversations that I always love and enjoy. I've had two medical students from the University of Chicago, Scott Goldberg and Annie Castro that work with me. And Randy Bollie, who's sitting over there is my current research assistant. And he took the time to put together that wonderful map of all the McLean Center Fellows, which we will give you a copy of. Hi, I enjoyed your presentation. I know I've already had two questions, but between the cyber ethic and then the development by many corporations to develop cyber dust, where you can just remove what you commented upon, your tweet that then can be taken up and used detrimentally against you. I'm concerned having learned from Dr. Christakis and look at obesity and addictive behaviors in the community, we're gonna lose those precious jewels that you talked about, the nodes and the influencers. So how can we straddle the identity protection and then extract, as we heard earlier, these human capital factors that will actually drive health and wellness? I mean, I think one of the things for us to consider is that the idea of stripping, identifying information has been brought up as a solution to some of this. And you're right. What happens is you strip that, the data on some level becomes protected, okay? But depending on how big the network is or how small the network is, the relative degree to which you can ensure that privacy is limited even with the stripping. But you're right, we lose those nodal individuals. But I think the field is moving. One of the things that we've learned is that social networks aren't static. They change, right? Those nodal individuals rise and fall in their power and their location almost as we're studying them, right? So what we have to do is actually study the factors that make them nodal, right? Then we can go out and recruit those individuals without necessarily using that identifying information from the network. Then we're looking at what are the flags that identify this person within it. And that's more enduring data in some sense because the network, by the time we captured, it's already gone, right? It's behind us, so yeah. So one more quick question and answer, Dan. Dan Brunner, U of C. Thanks a lot for a thoughtful talk, LaRange. This is really very topical. I don't know if you can really comment to this, but I've just recently, in the last hour, received an email urging me to join Doximity because of its influence on the U.S. News and World Report and top hospitals. And I feel a little bit oppressed by that email. The author is sitting in the corner there. And that I really don't want to become a member of Doximity. And can you comment on this move of using social networks in this way? Well, since I'm not a faculty member at the University of Chicago any longer, I think I could comment quite well. No, I understand that. But there's two layers to that, Dan, which are interesting. One, first of all, I don't think we should have to be on any social network, electronic social network that we don't feel will benefit us in some way, right? I mean, I went on Facebook for a time because I felt like I needed to get a feel for it in order to understand it. Because so many of the young people I work with use it. And now increasingly across the entire population use it. I actually work with a social media guy for the Road Home program. And he comes in and tells me, he's like, yeah, everybody thinks that the seniors aren't on there. He's like, they're all on there. He's like, this is the platform. But you need to choose, right? My criticism of this whole thing with US News and World Report is that that is an entirely flawed system for metricizing medical schools at the get-go. And our leadership has continued to place value on it. I got a similar email at Rush, encouraging me to go online and complete the survey and all this stuff. And so here we are all sort of tacitly endorsing this single publication with a very questionable set of metrics in terms of how they measure and rank medical schools. And secondly, should we really be ranking medical schools at all? Like is this a valued proposition when actually all of our medical schools do very different functions and should be doing different functions? So I guess my question back to the thing is because for a long time, actually Stanford actually opted out of that process and tried to very hard. The previous dean there, Phil Pizzo, actually fought hard against this. And he didn't get a lot of collateral support. But if the top 10 medical schools in this country get together and say, we're not playing this game, we're providing this data, from my perspective, I think that whole system starts to collapse. So maybe that's what you can engender, Dan. Thank you.