>> Eric Horvitz: It is an honor today to have Dr. Walter Greenleaf visiting us at Microsoft Research. Walter Greenleaf got his PhD in the area of neural and behavioral sciences at Stanford University, where I actually met him years ago, and Jerron as well has known him for many years. His current research focuses on product development efforts with use of simulation technology to improve neurological, cognitive and physical rehabilitation. He has over the years jumped from academic scholarship into the world of startups in this space. He is currently the founder of In World Solutions, a company that actually specializes in using immersive technologies, augmented reality, virtual worlds for behavioral healthcare. He also is CEO for Virtually Better, a company that helps virtual environments for the treatments of phobias, anxiety disorders, PTSD and even treating addictions. He is a pioneer in an area that was called, subfields called virtual reality in medicine or health care, lots of technologies and ideas that came out of that and even a conference or set of workshops. He is currently also a scientific advisor and proposal reviewer for the US Public health service, the NIH, NASA… >> Walter Greenleaf: Stop, stop. >> Eric Horvitz: Well, lots of advisory roles and he has been in the media and he has been communicating about the possibilities here for a long time. Walter, go for it. [applause] >> Walter Greenleaf: Thanks Eric. The minutia is too much. I really excited to be here and one of the reasons I am excited to be here is that I think that Microsoft is in a position because of some of the technology you guys are promulgating, to really make a difference in healthcare and in the part of healthcare that I have been working so hard on. And it is an important area. It is sort of a part of medicine that has been a little bit by the wayside, occupational therapy, physical therapy, psychiatry, psychology; they have all been under technologized over the years, for many reasons which I won't go into here. But with the Xbox, the Kinect system, with some of the virtual world technology you guys are pioneering, it is really making a big difference in my field. And I will go into detail about that. But I do have a favor to ask of my audience. There is a substantial amount of material that is interesting to cover and there's also enough time to cover it. But it is going to be a little drab and boring if it is just me giving a spiel, so I am going to say some things, but I want you guys to participate too. Give me some questions. Object, object fiercely, criticize, comment, expand upon so that our audience is not just listening to me give this spiel. Okay, let's start. Here are the four main points that I'm going to try to cover. It is not just, what I am going to talk about is not just something that recently sprung out of a clamshell; it is something that has been founded on more than 25 years of foundational research. And I will talk about some of the data, the FRMI studies, the longitudinal and latitudinal studies that have been done, but there is a foundation of research. I also want to talk about why it's an inflection point right now that the changes in technology and changes of cost, more importantly is making this available to people both domestically and worldwide. It is part of the breakthroughs that have been going on in telemedicine. I will give some examples of how virtual environments are being used for a whole variety of things, not just in behavioral medicine, but in physical medicine and as a disability solution. And then I will talk about some of the things that are just around the corner. Some of which are being pioneered right here At Microsoft Research. Actually a number of the things--I give this spiel to other people and probably at least five of the seven things I'm talking about that are just around the corner are being pioneered here at Microsoft Research. As you guys know virtual environments have been around for a while and they are predicated on some foundational technology that came out of research labs more than 30 years ago, with general purpose simulators being a recent invention. They used to be kind of a primary focus of virtual environments to be focused on one specific thing like a flight simulator. But this gentleman here, Jerron, was one of the first guys to promulgate the idea of having a general purpose simulation. Now this was back in the days of computers being, to do the simulation being $400,000-$500,000 on a silicon graphics machine. And Jerron I think one of the first times I think you elucidated the idea of using a virtual environment to do programming was in this article from Scientific American with the data glove using, to do coding. But this concept of having a general purpose simulation was very important and it sprung a number of people, including myself, getting involved. And for me it's been sort of an exciting journey. Here's a picture of me back when I was in my late 20s I guess, when I started on this endeavor to get involved in using virtual environments. What excited me was the idea of using synthetic environments for stroke rehabilitation. I had known a number of people who had had family members who had suffered a stroke and I really thought the virtual environment would be a fantastic way for us to be able to help manage the rehabilitation process. And finally with the advent of the technology that is present now we are seeing this technology coming out and being used not just in the laboratory but the technology is escaping the laboratory and making it into clinical care. Of course, in medicine one of the areas where virtual environments have made a substantial difference in our starting to become part of the standard of care, is in the 3-D visualization, preoperative planning and data fusion in the middle of a surgical procedure where a CAT scan or MRI image may be reviewed and used to plan a surgical operation and then overlaid upon the operating field to help guide someone. It is also used for training. So that surgeons can use the dexterous movements of new cools to be able to do a complex procedure. And also now we are starting to see sort of a separation between the end factor with the human and the surgical tool with robotic assisted surgery like the da Vinci system. Some people argue that there is a hype cycle that goes on with emerging technologies and they would suggest that virtual reality is right through the trough of disillusionment and it is starting to come out on the other edge to what is called the scope of enlightenment and the plateau of productivity. I am not sure where virtual reality really falls on this, but there sure were a lot of really bad movies back in the early ‘90s about virtual environments. >>: So right now cloud computing is the most overhyped thing. >> Walter Greenleaf: That's what these people who have made this graph would, I guess it's the Gartner Group, [laughter] well yeah, and e-book readers. You know, this is all in augmented reality is in the advancing slope of being overhyped. >>: Oh I see. >> Walter Greenleaf: My point is… >>: It doesn't help but that the 2009 picture, how much that picture has changed just [inaudible] if you dangle on the graph, things are all different now. >>: Tablets would seem to be main stream at this point not due to us unfortunately. It is interesting. It's a cool slide. Where did you get--Gartner, huh, amazing. >>: It's two years old now so I think it… >> Walter Greenleaf: You know let's for the sake of discussion white out all of the specific text here [laughter] and just say that at some point virtual environments are on this curve and I would like to think that the phrase is no longer, just like artificial intelligence at one point was a sort of an overhyped phrase, I think the phrase is now a useful phrase. When I talk to clinicians about using a virtual environment as part of the therapeutic process, they no longer look at me as if I'm crazy, and there is actually, we will go into this later. But there are several international societies of cyber psychology and society for international virtual rehabilitation. There are a lot of clinical journals and research groups. So the phrase has made the transition from being a sort of an overhyped phrase, a hyped concept. It spawned a number of lousy movies and a few bad television shows. But now it is on the other side. We are starting to see it as part of the tools that we use for clinical care. Right now some of the virtual environments that are out there are incredibly realistic. But they are missing something very important that will go into. They can be very photo realistic with real-world physics and adequate sensory immersion but one of the pioneering things that Microsoft Research is part of right now is the avatars don't often have the right nonverbal behavior, facial expressions, body language and movements. In order to seem real, the avatar--you guys have seen this in Second Life, I'm sure, where everybody is sort of walking around like this and they have pretty blank expressions. For my purposes for the use of, in clinical care, we need to have more natural body language, natural facial expressions. So even though there have been great breakthroughs in VOIP and better graphic processors, broadband, movement tracking, low-cost computers, all of this has really helped advance the field but some of the late breaking things that are just coming out now that are incredibly important are the use of nonverbal behaviors and gestures as part of the experience. Again to move along, there has been a foundation of more than 25 years of clinical research, but just now this technology is escaping the lab and making it into the clinical area. Let me talk a little bit from a neuroscience perspective. In order to change a behavior, you really need to change your brain. This is often, you will hear people talking about years of therapy, and it can take years, because right now--previously we really haven't had the tools to break down that change into small finite steps. Also, repetition is required and repetition can be boring. It is also important to involve the brain's reward systems with positive feedback in order to engender the change. And that is one of the things that has been lacking in both behavioral therapy and also in physical therapy and occupational therapy, getting that feedback loop and enclosing it with a rewards system, breaking things down into small steps. So a lot of what I am going to be talking about is going to involve how do we engage the human cognitively to participate in their own change and rehabilitation, in a way that is not boring, that activates the reward systems and allows someone to do something that is repetitive but with a little bit of change, and ability to monitor progress and get feedback. And I will show you how we do that. It is really critical to engage the brain's reward systems and here is a book that I just found out about; I have not read it yet, but I like the title, the Compass of Pleasure, How Our Brains Make Fatty Foods, Orgasm, Exercise, Drugs, Generosity, which is a social experience, Alcohol, Learning, Gambling Feel so Good. These are the sort of things in a gaming environment that we can leverage the brain's reward systems to motivate us to change how our brains work. And that is through the basic principle of using virtual environment in a therapeutic manner, is harnessing of the brains systems and the reward systems in particular to help engender change. Here are some examples, I am going to start first with behavioral health but later I will talk about physical medicine. Here are some specific examples of how virtual environments are being used clinically right now in behavioral medicine. We treat phobias and anxiety disorders, posttraumatic stress, developmental disabilities such as autism and Asperger's, conduct disorders, drug and alcohol abuse, anger, eating, impulsive disorders; the list goes on, and I will show you how we do it. Let's skip a little bit here. Here are just a few quick clips of some of the virtual environments that we are using therapeutically right now. You will notice that there is a whole variety of graphical resolution. There is no coherent platform right now and no standards for the virtual environments that are used in the behavioral environment. It is just sort of all over the map. Some are very cartoony, some are very photo realistic, but it seems to us, the researchers in this field, that the degree of realism per se, the graphical realism is not as important as the cognitive engagement, I will go into what is involved with that. So the examples here, there is a virtual liquor store that we use of helping people with problems with alcohol and substance abuse get familiar with what otherwise might trigger a craving and learn to master the response to that craving. I will show you examples how we use that in a social environment. Here is a World Trade Center virtual environment that is used to help the still large number of people that are suffering from posttraumatic stress from living near the World Trade Center or watching it on TV. There are still a large number of people that are suffering from the cognitive problem of that and years later you can still get help and make improvements. Actually I recently talked to someone who was a Vietnam vet who had been suffering from posttraumatic stress for I guess more than 40 years now. Recently he went into a center that used a virtual Vietnam to help him overcome his posttraumatic stress and despite years of therapy previously it was being able to use that virtual environment that helped him overcome his posttraumatic stress. I will go into the details about how that is done. There is another example here of using a virtual environment for what we call stress inoculation. The Center for Disease Control is using virtual environments to help prepare their first responders for disease outbreaks so that when they go to, if they have ever been camping before and all at once they are of in an African village and learning to deal with the jungle, you can prepare them cognitively so they don't have a stress reaction. Another example is using a virtual environment to help with fear of flying. For fear of public speaking we can use a virtual environment to reinforce-- it's amazing how well this works. To have an audience that actually smiles and claps when you are speaking, gives you positive reinforcement, can help you overcome any hesitation; this is your cue, audience. So the point is we can use virtual environment to change how you respond to a specific situation that otherwise might be very stressful. And again we use a virtual environment to help with the social cues that you need to master if you are having a substance or addiction issue like nicotine or alcohol or other addictive substances. In some cases we use actual green screened individuals to put into, as the avatars, mostly because we need to micro-expressions on facial expressions to do this. But in other cases we just use very graphic avatars. And I will show you the distinction of why we do that. So that is sort of my smorgasbord in behavioral medicine of some of the areas. >>: I have a question. In the previous slide what was the application of the people around the pizza boxes? >> Walter Greenleaf: That is for people that… >>: The other ones I get public speaking, flying, what is that one? >> Walter Greenleaf: That one is for, it is a party scene. And a clinician can use it for someone who is having trouble walking into a party, being offered a cigarette, social phobia. It can also be used for just if you have nicotine issues, alcohol issues. Often what happens to people is that they are very good at maybe at home not reaching for a cigarette or reaching for a beer, but they will go out to a social scene and sensory clues in the social clues will trigger the behavior that they are trying to avoid. So what the clinician will do they will have the person practice going to a party, and it works. You might not think it does, but it really does. You go to a virtual environment, you practice saying no to the beer that's offered to you. You do that over and over and over again, and eventually you get it down. Why can't you just do that using your imagination in the clinician's office where you are sitting there and saying okay, now imagine that you are going into a bar and somebody is offering you a beer. What are you going to do? You might think okay, I am not going to do it. Again unless we engage the brain’s--the more sensory systems we can involve, the better the response we get. And it can make the difference between years of therapy, years of counseling, versus just a few sessions. And to that effort we do have some olfactory stimulation systems that really seem to help a lot for some of the things that we are doing. I am going to pause for a second because I have given 10 minutes worth of the spiel here to say hey, what do you guys think so far? Is this lining up for you any questions at this point? Okay, all right. Let's push on then. I break down what is going on right now in therapeutic healthcare with virtual environments into sort of the four zones. There are the very immersive virtual environments where we use an HMD. You may have a tactile and olfactory stimulation that goes on hand-in-hand with that. These are typically used for the posttraumatic stress, the phobias and anxiety disorders. Another one is the interactive social virtual environments. These can be done with a flat screen. They can be done over a network. But they emphasize the social interaction between individuals and this is where we need to have the facial expressions and body language et cetera in order to have it work. These environments are used for addictions, for learning social skills, dealing with problems such as autism and Asperger's, anger management, impulsive disorders, attention disorders like ADHD. The next zones are the zones for health, for wellness and prevention, where you are trying to learn better eating habits, better exercise habits, smoking cessation and et cetera. These often are done on the flat screen game like environments which we described, but they have a completely different paradigm. The condition is not involved; these are often things that can be done on your own with the help of autonomous coach or an algorithm to help you manage your progress in the therapeutic procedure. And the final area is mostly dichotomized because it uses other sensors. And this is the zone of physical rehabilitation and injury recovery for stroke, traumatic brain injury, et cetera. Has anybody in this room ever gone through physical therapy from an injury? Okay it is really boring, isn't it? You have to have the personality of an athlete to do it. And one of the things that is coming through right now is we are using virtual environments to make it both social, you can do your rehabilitation over a network with other people, and see your progress and make it engaging. So no longer does it have to be something where it is painful and boring, it can be painful and interesting. Let me show you some examples of the immersive treatments that we are doing for posttraumatic stress et cetera. I should mention that our collaborators which I think folks here at Microsoft Research are probably doing some work with, Mark Bullis at IC2 is doing a lot of work in this area and Skip Rizzo is one of the early promulgators of using virtual environments for a variety of problems such as posttraumatic stress and cognitive rehabilitation. So I will go into detail a little bit on this one. There are a huge number of unfortunately servicemen and women coming back from extended tours of duty in combat zones and they may have a traumatic experience that is acute, where they are driving in the Humvee and there is an explosion and they see something very, very disturbing like a friend being killed. But there is also just a chronic stress of being in a war zone, constantly worrying about what is going to happen, being hyper vigilant for years. And this results in long-term disabilities. And I will give other examples from earthquake posttraumatic stress et cetera but we use combat as an example here. So consequence might be you come back, you are no longer in the military, you are driving on I-405, yet you see some trash on the side of the road and it causes a panic response. Or you go under a bridge and you are worried somebody is going to drop a grenade down on you. It is very hard to unlearn this learned fear response. And it can haunt you for years. In the example I gave earlier about someone from Vietnam suffering from problems 40 years later, yet, there is a way to treat it. Basically you have learned a limiting system fear response that is triggered by visual and auditory and olfactory stimuli and it is a survival response, a mammal to have that sort of instinctual reaction to something that almost killed you before. To unlearn it, you need to have, to go back, stimulate those parts of the brain that have learned that response. You do it in a controlled way so that what we call our higher functions start kicking in and saying oh, that piece of trash on the side of the road, I know what that is. I am no longer back in a war zone; it is just trash. It is very hard to do that if you're just using your imagination. Your brain, those are painful memories, painful images. And besides not all of us are very good at reimagining things anyhow, only a handful of people can really have the vivid visual and olfactory auditory memories anyhow. And your brain doesn't want to go there if you have had trauma over it. We can take people there in a graduated manner, do the exposure and immersion, cognitive immersion to habituate that feared response and it works. So here are some clips of the system being used. Currently we have recently deployed in 10 Air Force bases and I think we have probably 30 systems out at VA hospitals are now. It is a great success story. A clinician will work with a patient. They use an HMD. And they go through, and I will show you some video clips of this, they go through in a graduated manner. You might start with something that is non-traumatic at first and just get them used to the system, returning to the scene of the stress. And the counselor helps them manage their anxiety. They practice breathing and relaxation techniques. It may not take several sessions; it may take 10 sessions. But after a while, the brain gradually habituates and you can recover from posttraumatic stress. And again the same paradigm can map to fear of flying, fear of heights, early trauma of childhood abuse, et cetera. There are lots of stresses out there that cause those problems for people later in life. And the same paradigm can be used to help them overcome them. Jerron? >> Jerron: No go-ahead. >> Walter Greenleaf: Let me show you a clip on how we do it. This is from an Xbox game actually. [video begins] This is from Skip Rizzo's virtual environment, and a clinician might just start out with this or a more neutral scene where they are just driving along in a Humvee but nothing necessarily traumatic happens other than just been reexposed to the scenes of war. >>: The basic principle here is exposure with some notion of relaxation or distant seeing from the negative aspects again and again and again [inaudible] is a healthy thing to do to get beyond the fear of a single incident or a scary incident? Is that the principal? >> Walter Greenleaf: Yeah, you trying to reassociate those sights and sounds and smells with a more relaxed situation and you are trying to let people know that they can think about these things, be exposed to them, and not have the panic response. So in a way you can just simply describe it as habituation, it is called exposure therapy. >>: So as you expose, how are you relaxing? >> Walter Greenleaf: Well the clinician… >>: This is not exactly relaxing [inaudible] >> Walter Greenleaf: Right. Well the clinician is there with you in the room. You know it is not real. At any point you can take off the HMD. The clinician is instructing you to relax, to remember that it is not real. The clinician is having you practice breathing. If they see that you are getting a little bit overstressed, they will say, why don't we stop for a moment and think about this. And it works. The biggest problem we have is actually having people volunteering to be part of this. This is something they are very fearful of. And unfortunately in the military, too, there is a huge stigma about posttraumatic stress. It is getting better but my understanding from talking to clinicians is that once they get someone to say I have symptoms of posttraumatic stress. I am having suicidal thoughts. I am having a lot of stress and anxiety. There are processes like this we can put them through to help get them back. But because they might lose their job, because it is not macho to admit that you're having this, it is hard to get people to participate. And that brings up another point I want to make which is one of the advantages of using avatarbased therapy is that we can do anonymous counseling, and anonymous help. So maybe help people start taking those first few steps where they don't actually have to go to the clinic, where they can show up as a virtual avatar. And it is not just for posttraumatic stress. If you have a sexual addiction, if you have a drug or alcohol or substance abuse problem, you might want to show up for your sessions to start the process as an anonymous avatar. And I will go into that a little bit later. >>: I was just going to mention when these applications of treating PTSD with virtual reality first were proposed, I was one of the biggest skeptics. I thought it was just too cute and gimmicky and what persuaded me about it were really just the clinical results. It was one of those things where I really came to it with a biased feeling that this couldn't possibly be for real and it turned out to be so, I was actually chastened. >> Walter Greenleaf: Well I will show you some of the data if we have time. It actually is pretty amazing. You can see a nice stepwise improvement that you don't see when you look at Wait List and Controls or you look at other ways of doing it. Now I should point out for some of these things like fear of flying, exposure therapy has been done for years. But they would actually have to take someone to the airport. Like for fear of heights they would actually have to take someone to a bridge, or the elevator and they would have to go with them, very expensive, very time-consuming. And for combat or an earthquake type of posttraumatic stress, it is very hard to take someone to re-experience that in a controlled manner. So the virtual environments give us the ability to do something costeffectively, possibly over the internet, and in a controlled manner with measurable data. And that is one of the keys to some of these things I am going to be talking about too is not only are we getting new tools for therapy, we are getting new tools for monitoring and measuring progress. And one of the reasons cognitive behavioral therapy, physical therapy and occupational therapy have been sort of under technologized aspects of medicine, is that they have not had good measurement tools. So one of the ultra messages I would like to portray here is that the technology that is been worked on here and other places is going to bring out a new ability to measure and index progress in a part of medicine that has largely been analog and subjective. These rich environments can be used to measure objectively things that so far have been subjective, but that is for another part of the talk. I am going to push on because I have a lot to go over and not a lot of time. So, very quickly, research has shown that it is more effective than just plain imagine exposure. It is as effective as the in vivo exposure and of course it is more cost-effective. You do not have to have an exact replica of the environment to have a therapeutic effect. You just need to have enough cues to stimulate that sort of anxiety response. But it does not supplant the condition; it is just a tool that allows them to do things that otherwise are too difficult or impossible to do. I am going to skip ahead to some of the work on fear of flying. This is an area where there has been a lot of research showing the clinical mechanism and validating it. And of course now we are starting to have much more high-resolution airports that we can take people to put them through. Here is an example of one of the studies on fear of flying. On the left hand of the graph you see virtual reality exposure therapy both pre-treatment, there is an index of fear of flying. That is a general scale, the post successions of treatment and then the six months follow-up. So with the virtual reality exposure you see a nice stepwise decline. Comparable to the standard exposure, but with the Wait List and Control they don't get better over time; they actually get worse. And this is a typical pattern we see for other areas where there has been research. Here is a study on Virtual Rack which was done at Camp Pendleton with the Marines that are coming back. Same type of progress that pretreatment, high levels of measurements of posttraumatic stress including indexes of anxiety and depression. Post treatment there is a decrease and then with a three-month follow-up, there is even more of a decrease. And it shows up not just in the advert symptoms of posttraumatic stress, but other measurements of anxiety and depression. We have looked at where is this going on in the brain, looked at it with interest with fMRI that mapped to neural plastic changes. And we can track the changes that go on looking at the MIgda activation, pre-and post treatment to using this approach. So this is not just a subjective index of improvement that we are getting from self-reports that we are getting from the patients going through but we are getting a subjective measurable changes correlating with the improvement. And there've been a number of meta-analysis that has looked at the value of this for treating exposure therapy. I should mention though that despite the clinical validation studies have been done, despite the almost 20 year’s worth of research in laboratories on this, it is just now making the transition from the universe of laboratory to clinical care. There have been barriers to entry on a market basis. A lot of which has been cost. And that is one of the reasons I am excited to be talking to this group is that the Xbox, the Kinect and some of the other technologies that you guys are pioneering are making this affordable and that will make the difference. If it is just an expensive way of having a therapeutic approach, the clinicians won't be able to take the time or afford to do it but making it affordable makes all the difference. Okay let's talk about just generally. I was talking about one specific area, posttraumatic stress. But let's talk about a general cognitive behavioral therapy. I won't ask how many people in the audience have had to go through counseling or psychiatric or psychological care, but I will point out that for most people it is a long drawn out process that is often very low yield. I am sure the clinicians out there that will probably take issue with it but it takes time and it doesn't always work. And part of it is, in my opinion, is that you have to recall past events. You have to convey them verbally to the clinician you are working with. You have to do some role-playing and imaginative work on how you are going to handle the situation. Maybe you are dealing with getting into a fight with your spouse all the time or maybe you are dealing with an anger issue that you have. Or maybe you are dealing with an addiction that you have. It is very hard to just talk about these things. But if we use the virtual environment as scaffolding for these discussions, it can make a big difference. I will show you how we use it. Again with what goes on with just talk therapy, it is very hard to change the neural system behind the behavior because you are not activating enough parts of the brain, at least that is my biased perspective on it. I feel that in order to change the brain system, you have to activate it. And if you're just using talk therapy alone, you are not engaging the patient and activating enough parts of the brain to do it. So what we are using right now, interactive virtual environments in a variety of situations. I will show you a few examples. Now in this case we are stepping away from using the fully immersive, with 3-D audio and 3-D visual and sometimes olfactory stimulus to just using a standard keyboard and visual display. Despite the less immersion we are finding that it works very well especially with younger patients. For example, we are using it right now at a clinic in Newark New Jersey with kids who are 14, 15 years old, both girls and boys who are in a locked residential care facility because of their behavioral problems, conduct disorders, violence. These are kids that are there because people believe that they don't have to spend o the rest of their lives in prison. But unless they learn how to control their impulses, these are kids who have killed other kids, or other people. And these are kids who burn down buildings and I won't go into lists but there is hope. We are using the virtual environments and the clinicians tell me that it is working to for them to learn to control their impulses and modify their behavior. So in this case we use a 3-D interactive virtual environment to do role-playing and interaction. So the clinician may take on the role of another teen and they may discuss or do some role-playing as to how to deal with a very frustrating situation. And they will practice how to respond to it. Now the clinicians tell me that if they just sit down with the kid and try to do some role-playing in two chairs sitting and talking, these kids won't stand for it. They will pick up the chair and throw it across the room. They will not pay attention. They get maybe 5 minutes at most of engagement of these kids. But they will sit down and spend an hour in the virtual environment. Part of it is because they are little bit more anonymous. Part of it is that they don't have to worry about what their body is doing. Part of it is that the clinician is no longer the man, an older adult; he's taken the role of a teenager. And so we have had great progress in having these kids learn to control their impulses. That is just one example. I will give you some examples further downstream in my talk about how it is being used for kids with autism and Asperger's to learn social skills. But these flat screen interactive virtual environments have proven to be some breakthrough ways of helping reach people with cognitive behavioral problems that otherwise cannot be reached through therapy. The other advantage to using a virtual environment for cognitive behavioral therapy and other forms of rehabilitation is you can create a community. Often the therapy is in isolation, but if you are doing family therapy if you are doing things that involve, for example, a service man our woman coming back with an amputation, what we are doing right now is we are working with the Air Force to help people who are deployed stay in touch with their families through a virtual environment where it is not just videoconferencing where your are doing postage stamp look at somebody else and you're talking with them. If you have a five-year-old kid this videoconferencing only works for a few minutes. They want to do something with you. In a virtual environment you can go for a ride, you can go for a walk and you can go play a game or you can go for a scavenger hunt. Then later if there is an injury and dad is returning and he is making his progress through the health care system before he is coming back, a clinician can meet with the kids in the family and say look mom’s coming back, dad’s coming back, he's going to be a little bit different. He had a brain injury or he lost an arm, prepare them psychologically and also prepare the serviceman or woman of what is going to be like to be--we can build a community and we can help people make transitions through the use of the virtual environment. I'm going to pause for a second because I have been talking uninterrupted for a while, what do you guys think so far? Any questions? >>: One question I have is in counseling have people thought about, I am curious about various studies that integrate the fidelity in characteristic ways and see what the loser gained [inaudible] interactive flat screen presentation of [inaudible] 3-D world and you go to really beautifully rendered sets of pages of scenes, how much does that get you to [inaudible] versus Norman Rockwell homecomings for service members [inaudible] there must be some effect [inaudible] >> Walter Greenleaf: I will tell you a couple of thoughts. I think there have been some studies mostly looking at the difference of HMDs with 3-D sound versus flat screen. But the need has been so acute and the funding has been so miniscule that there really hasn't been a lot of research in this area. I can tell you some of the clinical impressions I get from people who use the virtual environments. For things I do like the phobias, the posttraumatic stress, a little bit of a clue goes a long way. People's brains take them there if they just get a few little stimulus. It doesn't matter how high resolution it is. For the adults, I mean older adults, let's say, who are using virtual environments, they haven't spent a lot of time playing video games. At least not most of them, so it takes him a little while to get familiar with it. But there the fidelity doesn't make a big difference. But for some of the teenagers that we are working with, if you put them in a bogus virtual environment which is just like lame like two-year-old technology they don't want to participate because it's old. So I think in a way you are asking two questions, Eric. One is what is the level of fidelity we need to get a clinical effect? And two is sort of the social context for doing this. What is going to make somebody want to participate and comply with participating in it. >>: [inaudible] I’m not sure if I should ask this or even bring it up, but it seems kind of scary thinking that there is a positive therapeutic effect in going about the [inaudible] version of warfare [inaudible] modern teenager [inaudible] >> Walter Greenleaf: It's important question and I am glad you brought it up. For years some of the videogame people have been saying, you know, these games that have you stabbing, shooting and blowing up things all day long, some kids will spend 12 hours a day doing this, doesn't transfer over. They don't learn from it. But now that we are starting to have some healthy games come out that are using these virtual environments, some of the video companies are saying, you know, it can transfer over in some contexts. You're absolutely right. I think that we do reprogram our brains all the time by how we spend our time and what the context is. And I would like to see that kept in mind as games are designed. I think that that would be important. My role is to try to leverage all this game technology and all this military simulation technology and move it over for therapeutic purposes. But if you're asking me do I think some negative things can happen because of being immersed in violent video games, sure. Yes, it does happen. Yes? >>: You see a future for Xbox is that there is a new conferencing app called [inaudible] avatar. Playing in this… >> Walter Greenleaf: Avatar Kinect. >>: You have to worry about the [inaudible] because they are all avatars. >> Walter Greenleaf: Right. I have got that a little bit further down in the spiel here. But yes, it is going to make a big difference in what we do. >>: Especially if there's physical attributes to the therapy and may join in groups. That might seem like it would help. >> Walter Greenleaf: Absolutely. For group therapy, for networks therapy, sometimes having an anonymous presence, especially--I'm finding that for the work groups that we do for kids with autism and Asperger's who are learning social skills, somehow not having to worry about what your body is doing, somehow not having to make eye contact, that puts, and also having control of the world to a degree and complexity allows the clinician or counselors working with them to staircase them up into learning some social skills otherwise that they might not be able to do. There are some really great things we can do with the technology that you just described. For example part of the miss-development in the social brain that often happens with autism and Asperger's is the ability to recognize correctly facial expressions and other nonverbal clues such as body language and gestures. We can exaggerate those in a virtual environment and have someone practice recognizing them until they get it. And then we can reduce the exaggeration until they can recognize it with a normal level of realism. So we can finally take social interaction and adjust the degree of fidelity for therapeutic purposes. And it is really going to make a big difference, and the ability to measure movements, well that has great implications not just for the therapeutic aspects and behavior but also for stroke rehabilitation and traumatic brain injury rehabilitation et cetera. I will get to that in a moment too. Let's keep moving unless there are any other questions right now. Usually at this point I would invite my colleague Yvonna Steigman who is out there doing a lot of clinical care research right now to join us. This is, but she is not available right now. But this is the user interface that we have developed for using one virtual environment for online interactive therapy. The avatar is pretty good. If she was moving around and talking with us and giving a tour right now you would see that she has this natural bio kinetics and movement is pretty natural. There are great facial expressions which are linked to the tone of her voice. There is lip-synching so when she talks her lips move and the facial expression maps some aspects to the tone of her voice. We have buttons the clinician can use to not just give certain gestures but to convey a motion. There is another set of gestures that we can pull up for different emotions like happiness, sadness et cetera. So Yvonna uses this as a therapeutic environment either co-located where she is sitting on one computer and the person she is working with the sitting on another, or for group therapy where they are all sitting together on computers in one room but meeting in the same virtual environment, or distributed across a distance. An example is a group of social skill work with for kids with Asperger's, where they will meet Wednesday nights at the clinic and they will all participate in the virtual environment at the clinic. But then on Thursday night and Friday night they meet from their homes and practice the socials skills learned in a moderated group environment. Part of the problem we have with this virtual environment right now is exactly what I described. We use a button to say happiness, a button to say anger. So we convey the emotions to the avatars here in a very binary manner without very much finesse. So one of the things I am very excited about with the Kinect technology is to be able to map facial expressions to our avatars. I can't wait until we can do that. We don't have it in a clinical environment yet, but you have created the underlying technology for that. So this is the virtual environment that we are using right now for substance abuse issues, for social skills training, for anger management, for addressing issues of family therapy, a whole variety. And I will show you some examples of that. Here is an example of using this for family therapy, teaching somebody in a group therapy situation how to deal with a family issue. Here is an example of [video begins] >>: Okay [inaudible] are you ready to role-play? I will play your mother and you can just be yourself, right? >>: I think so. >>: Okay let's all report to the house while I change my avatar to look more like Debow’s mother. >> Walter Greenleaf: The wallpaper is horrible in this example. >>: So son, what did you want to talk about? >>: I am not really sure, but I know that I let you and dad down and became a great disappointment. >> Walter Greenleaf: One advantage of doing some of the role-playing in a virtual environment is you can capture the interaction and then review it for sort of a debriefing. So that was an example of how they might have done some role-playing of what the son is going to do to talk about it to his mom about a difficult subject. But later after doing the role-playing you can save it on smart phone or on some other pad like device to review and practice. If you are dealing with how to deal with a bully at school, or if you are dealing with how to deal with a difficult coworker, you can do some role-playing and then review it later. I will give some examples of some social skills work too. [video begins again] >>: Okay, thank you all for teleporting to the school. >> Walter Greenleaf: Of this is from role-playing that we did with a social skills group with kids with autism. >>: And today your [inaudible] is inviting your friends to the movies. Remember how we were learning earlier this morning in our circle about how to start and maintain a conversation and how to end a conversation? So I would like to know who would like to volunteer to practice those skills. >>: I would. >>: I would. >>: I will. >>: Okay, how about we have Latoya and Linda. If you guys could all, the rest of the kids if you could all circle around and let them be the center so they can practice their skills. And please remember that while we are practicing here that we are going to give feedback to you and this is only to help you not to hurt or make fun of anybody. So I just want you to know that we are all here to support you, okay? So go ahead, let's start a conversation. >> Latoya: Hey what's up? >> Teacher: How about their body space? Remember we talked about body space? >> Walter Greenleaf: Okay you get the idea. Go ahead Eric. >> Eric Horvitz: Are people interacting live or if so, what are the controls they have? >> Walter Greenleaf: It was that user interface that I showed you earlier. Moving the avatars is by mouse and keyboard movement and that is why they look sort of jerky. >> Eric Horvitz: So they are trained in navigation [inaudible] >> Walter Greenleaf: Yeah, well with kids, you don't even have to tell them. They just sit down and they know that the W in the S and the keyboard are for certain movements, you know, old-school. So in that case, it was that paradigm that I described before where they are all meeting at a clinic. They all have their own computer. They are meeting in a virtual environment. And then later they can meet off-line from their home to practice with the moderated sort of chat room way of practicing social skills. >>: What type of platform are these examples running on? >> Walter Greenleaf: That was from Olive. >>: That was Olive, okay. >> Walter Greenleaf: Since we are online interactive virtual environment it is 4 Terra, out of San Mateo developed it; it was recently acquired by SAIC. We use it because it's HIPPA compliant, which is very important for what we are doing. These become medical records, which is another whole issue of discussion we should talk about. But it is important to be able to, you know, keep the medical record confidential and the voice is identifying for the individual. One could argue also that the body language, once we start mapping back, and the facial expressions that is going to become uniquely identifiable too. So you have to treat these as medical records even if they are in anonymous avatar, if you are capturing the person's behavior, you are making a medical record. So it is something to keep in mind. Okay here is just one more quick example. I want to push onto some other things. This is like using this for an anonymous AA meeting. [video begins] >>: Good morning everyone and welcome to the [inaudible]. My name is Barry and I am an alcoholic and I chair this group. >>: Good morning Barry. >>: Hi Barry. >>: Good morning Barry. >> Barry: Good morning everyone. The focus of this meeting is to share our strengths and help each other so we can solve our common problems with alcohol addiction. Let's go around the room and introduce ourselves… >> Walter Greenleaf: So it takes on the format of a standard clinical therapeutic environment. It doesn't have to be the usual AA structure. But it allows people to A, participate in a telemedicine manner and that is really important to keep costs down. And B, for people who might be shy about talking about a psychological problem, it breaks the barrier when they can show up as anonymous avatar and get help. It is going to open up therapy to people who otherwise wouldn't through cost, distance barriers or their own hesitation to take that first step. They will be able to start participating in the therapeutic process. I didn't bring statistics. I'm sure you guys have already heard about it about the percent of mental illness problems in our culture. But it is legion. And as we are getting better with dealing with disease and the physical aspects of aging and so on the cognitive, the behavioral and especially in the stress involved world will start to become more and more large. The technology that is here is just coming out in time. I think it is going to be really important especially in a telemedicine manner to get it out to people. We will talk about that later. I will skip forward a little bit. We also use virtual environments for what we described as refusal skills training, practicing saying no. So we have virtual environments for, you know, someone comes up and asks you for your drink order, or you go into a liquor store, we have--For a clinician to induce a craving for someone who has an addiction problem is a big challenge, because you can't exactly show them marijuana. You can exactly show them cocaine in a therapeutic environment; it is illegal. Yet you need in order to get them to master their cravings, you need to stimulate that response. So we can do it very effectively with virtual queues. Again the telemedicine model makes a big difference. One of the things I am really excited about is behavioral medicine is really devoid of a lot of objective data. It is a very subjective world right now with very subjective measurement scales that are often just collected in an analog manner. Data piles up in files but it is never really--algorithms aren't optimized. While using a virtual environment in a telemedicine manner we can start collecting and anomatizing data at the backend to start defining protocols and this is really important. >>: That little thing you put on the left there that's a Kinect? >> Walter Greenleaf: You're absolutely right. It is an early version of a Kinect. And the snowman, who knows where that came from? Okay, this is an illustration that I did 20 years ago. It is just now coming out. This is a really important thing. And here's something else I wanted to talk about. The tools that clinicians have right now in behavioral medicine for measuring psychological state are limited. You may not have taken some of these tests. You probably have taken intelligence tests; I'm sure the people in this room have. Some of the psychological inventories like the MMPI test et cetera are very old. The answers are out there on the internet. So if you are using them to screen for employee issues et cetera, the answers can be found, people can break the system. So there is a need to come up with some better ways of measuring mental status, mental state that are ecologically correct and are variable. So what I want to do is use virtual environments as a better way of doing psychological assessment. Instead of a subjective assessment, to have some trials. You know, we have somebody be challenged. Maybe we have them, if we are looking for an attention issue, attention deficit disorder, why don't we have them go for virtual bicycle ride and have a bunch of distractions come up, and how you respond to the distractions, and how long attend to them, how quick can you get back on task? But do it in a functional manner, rather than the silly tests that we are doing right now. For more complex cognition executive functions, same thing, as we age we are going to want--there is going to be more therapeutic options out there. But unless we have the underlying data to optimize protocols, define the drug and therapeutic interventions, we are not going to be able to refine the response. So this stuff is really important in terms of improving the next generation of behavioral healthcare, because we will finally have data, data that we have not had before. Okay I am going to talk little bit about, there is a business model behind this. We can show how it increases revenue; how it reduces cost, how there are a number of value points for state and local governments. I think I remember reading recently that the state of New Jersey has spent 2 1/2 billion dollars on mental healthcare services alone, in one year. Mental health care and some of the subsidiary problems like violence and addictions and so on is extremely expensive. So the state and local governments have an interest in this. They are already spending a lot of money on not very effective therapeutic programs such as prisons. But patients themselves have an interest because it is less boring. There are better results. There is reduced stigma. There is a business model for the clinicians and the practice owners. So later if you guys have questions about the business model behind this I can go into that in detail too. Another example of where we are using virtual environments this is a very beautiful game environment that Michael created using for kids going through painful procedures like recovering from burns or having a bone set et cetera. So we create an engaging environment that is specifically designed to distract them during a painful procedure. It is just sort of a modest example of how this technology can be applied to a particular problem but it is one that is finally getting out there. And it is nice to have the graphical tools to be able to finally do this. And in this case we have a controlled breathing that goes along with the distraction. And in order to progress through the virtual environment you have to regulate your breathing. And that helps distract them from the pain. We are also using virtual environments for stress inoculation. This is a project we are doing with the Centers for Disease Control, to help them train first responders how to deal with going into an earthquake zone. I would also like to use the same environment for the posttraumatic stress from earthquakes. As you guys know there is Haiti, there's China, there is Japan with the tsunami. There is a huge amount of posttraumatic stress that will go untreated unless we can come up with a cost-effective way to do it. So again we use the virtual environments and with the same paradigm that I showed you before we do the therapeutic process. But you can also prepare people who are likely to go through this stressful process by having them go through it in a virtual environment and be prepared cognitively for it. It helps reduce the risk of posttraumatic stress. Here's another interesting application. This is training a clinician on how to present a terminal diagnosis. You would be surprised, and Eric maybe you remember this from your medical training, you would be surprised at how ill-prepared clinicians can be for delivering bad news. And if they don't deliver it very well, it can be a horrible experience for the patient. To make what otherwise could be a manageable issue become very difficult to manage if the presentation of the problem is not handled well. So that we are working with MD Anderson clinic, we have developed this system here for the clinicians to practice how to deal, sort of like a flight simulator, a flight simulator for delivering a terminal diagnosis. In this case we are using a green screen to deliver [video begins] >>: Hi. It is nice to meet you. I hope my results are ready. I was expecting to see someone else. >> Walter Greenleaf: So the user is typing in commands here. I would love to use some of the NLP work that has been done here et cetera to help refine this. [video continues] >>: I came alone. No one will be joining us. >> Walter Greenleaf: So there are a variety of responses that can be engendered. And angered response, a tearful response, a violent response, but the clinician can practice and prepare themselves on how to do this in an empathic way. Some of the users have been using this to teach something that is very important but very hard to train somebody on which is how to be empathetic to a situation like this. >>: [inaudible] there are two physicians in Seattle; they are the two NW. referrals for Lou Gehrig's disease and they work together to break it to the patients slowly even when they are both certain that. Lou Gehrig's is easy to diagnose, well were not sure here exactly, but it's something that is not great. I will tell you what wanted to go to and so they send them for a second opinion and they actually coordinate completely on the phone behind the scenes on how to break the news. It is such a difficult diagnosis for the family and they try to [inaudible] in the right places and so on. >> Walter Greenleaf: Well that is good that they have worked it out. From what I hear from my lot of the medical schools that often people just sort of get thrown out of there without a real chance to learn how to do this difficult thing which is part of being a good clinician. >>: [inaudible] experience I asked the medical students in the medical students to tell them, the young mother of a new baby that was having [inaudible] problems. They discovered that the mystery loss of platelets was caused by HIV and to tell her mother that her kid had HIV and by the way so did she. And it was all [inaudible] shock and had to be delivered in parallel with the family and the medical student on the case [inaudible] of doing this was [inaudible] >> Walter Greenleaf: Sure. And it is the sort of thing that you don't necessarily want to have on-the-job training. You want to know how to do it right the first time. Virtual environments have been used to train people how to do spinal taps effectively the first time so that they are not practicing on a real human. Now we can finally use it to teach some of the… >>: [inaudible] spinal tap [inaudible] see one, do one, teach one. Is the standard model in medical school on the [inaudible] patient. >> Walter Greenleaf: Right. And I don't want to be the first do one. Anyhow, one of the things that I think is going to be great is what is going on with using virtual avatars now is that it is expensive to do the green screen. It is hard for us to come up with a variety of therapeutic teaching environments, but we do it to get the micro-expressions of the face. So with progress of the facial recognition algorithms that you guys have, I think that we can eventually get that onto, so that we don't have to use the green screening which is very expensive. Just quickly here is an example of using a virtual environment for leadership training. [video begins] >>: The scenario that we are going to show you here is a reduction in force exercise so we gathered managers together from a big bank which has just announced its earnings for the most recent quarter which unfortunately are… >> Walter Greenleaf: Reduction in force means firing. So this is using a virtual environment to teach someone how to correctly lay off someone in a nontraumatic way. >>: [inaudible] that we can just donate to this project. [laughter] >> Walter Greenleaf: Well anyhow. This has been used by the business school at UC San Diego as part of their executive coaching is teach managers how to humanely hire and fire people and so on. Why would you use a virtual environment to do this? Well, it is less expensive. You can practice it off-line. And you can keep notes on how to do it right. And more importantly, and again this gets back to how do brains learn, is you can practice instead of just doing it on-the-job. I'm not going to go into detail on this one. I can show it to anybody who is curious later. Okay some of the work we are doing with ICT right now is using virtual environments as cognitive assessment tests. This is for being able to, a next generation of mental status tests for the Army. And again using the virtual environments instead of using paper and pencil tests to index memory and recall, why not use a virtual environment where you are asked to walk down an alleyway and recall what you see in there. And later you might be asked did you see a car? There is a lot of data you can get by just looking at how someone navigates and how they respond to stimuli and what their recall is. It is much more effective than the analog tests that are currently being done. They are hoping to use this to get some baselines from people before they are deployed so they can do an inventory after they come back to see if there has been an impairment because of concussion et cetera. Okay I am about to segue over, this has been a long spiel. I hope you guys are doing okay. I am about to segue over to talking about this for physical therapy, occupational therapy et cetera. But before I leave cognitive behavioral stuff are there any questions about that? Sure. >>: I was wondering what kind of [inaudible] the person would get when they're using the virtual environment. So for example if they are practicing in the real world if there are other people that can teach them about their mistakes versus when they are in the virtual environment. So what kind of [inaudible]? >> Walter Greenleaf: Well in the virtual environment they can get the same sort of feedback. If you are doing group therapy in the virtual environment, there are other real people there with you. And the clinician is there with you too. So they, it is the same level of group or expert feedback that you can get too. We would like to move some of this off-line so that you could learn some of these, practice some of these things like anger management for example, practice it on your own with a virtual avatar. And I will show you examples, some simulations that are being used for that. But we are not at that level of sophistication yet. Right now the interaction we have involves real humans interacting with real humans. We just use the virtual environments as a scaffolding to improve the therapy process. Outside audience, any questions? Okay. I asked you guys earlier if you've gone through physical therapy, you know it is pretty boring and often the tasks that they have you do are sort of simulations of functional things. It's like you might in order to practice, you know, manual movement they might have used stack colored rings on a pole. You might have stretchy bandages that you pull. One of the big problems is that it is boring of course, but also when you get home you are often sent with a Xerox sheet of paper with the exercises that you are supposed to do with maybe some stick figures showing you what to do. You often forget how to do it right or you don't quite understand the figures on how to do it right or you're not moving the right way and you end up doing damage. And there is no feedback loop. You come back to the clinic a week and a half later or two weeks later, you have been overdoing the exercise or under doing the exercise or doing it the wrong way. The clinician can only get several sample points and they are often just very subjective sample points on how you're doing. So it is very hard to come up with a trajectory for rehabilitation. It is very hard to dynamically adjust that protocol. So what is great about virtual environments is that we can make it less boring, more engaging, we can do it in a telemedicine manner so that we can get feedback on a daily basis on progress, send an alert to the clinician if the patient is not doing it right or send an alert to the clinician if they are making progress and the protocol needs to be stepped up a bit. Let me show you some examples. The stuff that is used for rehabilitation right now is usually pretty old and boring and not very engaging and again results are mixed. People can have an identical injury and someone might recover perfectly with one month worth of work and another person might never recover function. And the difference seems to be understanding what the process is and complying with it and that is something that is very difficult to do without the proper guidance. And we can use the virtual environments to provide the guidance. This is the study that I did with Stanford Hand Clinic. We took the people who were recovering from upper extremity injury and we hooked them up to a data glove and some other measurement tools. We hooked up some exercise programs that gave feedback in real-time. But more importantly sent the data to a server and allowed the clinician to check in and see how the patient was doing on a basis and get an alert if they had fallen off the correct protocol. This is something that we did about 10 years ago and again it is a tribute as to how hard it can be to get things once they have been proven in the lab out to the clinical community. We have still to get this out and commercialized. But it is a good example of the right way to do things. We set up an online chat group so that people could play an interactive game, and we found that with some of the elderly people that were going through recovery that they liked doing things that were more like a group like swimming together as you did your exercises as opposed to shooting aliens. And what was important was to be able to have them get feedback with a little video of what to do for their exercise, give them some real-time feedback of what their progress is and how they are doing, allow them to see their progress over time so they are motivated and again close that feedback loop so that the clinician can see what they are doing. And this allows you to have a tele-rehabilitation system too where if you are doing remote therapy, trying to reach underserved areas, it allows you to do it. Some of the big breakthroughs in, you know people collect rehabilitation now; I would like to change that to connectabilitation. [laughter] a lot of people are using--now this is actually a problem, as far as I'm concerned. A lot of clinicians are so eager to motivate their patients that they will say go home and use the Wii to do some exercises. And here is a game. But there is no managed protocol. It gets them moving, you know, they might move their upper extremity, but you really want to manage the protocol. And my friends who are doing research in this area say that often just sending somebody home with a game to motivate them does more harm than good. You really want to have a managed protocol where you are doing the right exercises the right amount, not overdoing it, not under doing it. So the answer is to have something that is designed to be a therapeutic tool rather than just sending somebody home with a game to do the process. Okay. I am going to talk about some of the things that are just around the corner, a lot of the things that are being done here. Well actually, before I do that, does anybody have any questions about--I skipped over pretty fast using virtual environments for physical therapy and stroke rehab et cetera, any questions about that? It is pretty self apparent that the value of the technology here. And trust me guys. We want this technology to come out because in 10 or 15 years when we are older and we might have a cognitive injury or a stroke or something like that, you don't want to be using those stretchy bandages for your rehabilitation. It is boring. And the good news is especially with some of the new pharmacology that is coming out that will help with rehabilitation that recovery from even a pretty significant injury is possible. But it is painful, it is hard, and it is slow. We want to have systems like this to help us do it, to help our parents do it, to help our friends do it, and by the time we need it. Here are some of the things that I am excited about. Here is something that I think I grabbed from Microsoft Research thing using texture mapping of faces to avatars. [video begins] >>: [inaudible] demonstrating a new approach towards making a photorealistic avatar using a regular 2-D video of someone as a reference and a simple 3-D model that can be created with Kinect. The research team synthesized a 3-D avatar event can be animated in real-time. This process of pasting on a 3-D video reduces the computation required for photorealism in real-time. As a 3-D talking head, I have many useful applications. >> Walter Greenleaf: I hope whoever is behind this is listening, because this is really great. This is really going to help what I do and my colleagues too so much, to be able to have the real-time texture mapping and facial mapping and the voice synthesis. This is so important. And something else I am excited about here. This is something with ICT. I think Microsoft is probably sponsoring some of this research too. This is the work of using simulations that serve the evolution of Klipee to be able to guide people through the morass of website information that the military has out there for helping families and people with posttraumatic stress. There is a huge amount of information out there. What they want to do is have a coach that you will bond with and trust a little bit, even though it is an avatar, even though it is a computer AI to help guide you to start the process of getting help for your problems with suicidality or your problems with substance abuse et cetera. So let's see if I can find it here. [video begins] >>: Well it sounds like you are experiencing some signs of depression. You know, things just aren't as satisfying as they used to be. Now this is pretty common for people who are going through major changes. And I know it is rough, but you need to be easy on yourself. Here are three things that can help, having someone to talk to, reading more about this thing you're going through, and getting back into the swing of things. Here are some links to get you started. >> Walter Greenleaf: So this character, they have done a lot of social engineering trying to figure out who is going to look comfortable for, you know, military servicemen and women to share information with. They have come up with, you know, other ethnic groups and other genders in case you choose to relate to that person. But this was, for example, a DARPA initiated project called Healing Heroes. Where they are going to build a social media site, I think that is not the operative name anymore, to study and provide the analysis of what people are typing as they interact to help guide them to find the right information. And it is done in a bit of an anonymous coaching type of manner where you get people to actually start sharing information with somebody that otherwise, you would never share any information with, a computer. I want to show you some other examples here. You guys have seen this one I am sure. This is extremely exciting for us. [video begins] >>: [inaudible] your voice and track your body. >> Walter Greenleaf: You guys have heard this 100 times. >>: [inaudible] facial expressions. As you can see now can track features like your smile, your laugh, and even the raise of your eyebrows. >> Walter Greenleaf: Since you guys have heard it 100 times I am going to cut it short. But that is incredibly important to what I'm doing, to be able to capture the facial expressions, to be able to capture the movements, to bring that onto an avatar. It is like the fifth dimension of interactivity and believability that we need for the type of care that we are doing. So it is very important and I am very excited about it. But that is part of my standard spiel that I give to audiences about what is just around the corner. Keep it coming please. Let's see here. Oh, some of the other things that we are working on now. We don't have things like hugging and shaking hands and crying and in the standard video games in virtual environments that are out there, there aren't. They are mostly pretty perfunctory in terms of being able to drive a car or been able to stab someone or being able to blow up something. So one of the things that we are putting into our virtual environments are more social things such as being able to have the bio kinetics right for hugging, for shaking hands, things, things that are really important and not part of the standard repertoire that we are seeing in virtual worlds now. So whoever is out there listing we want to encourage you to make sure that you build it into your virtual worlds all aspects of human behavior, because for therapeutic purposes we need it. Again for the realistic avatars, one of the things I am starting to see are algorithms for cultural specific nonverbal behaviors. For example, you know, in many of the virtual environments we are using for therapy right now, I notice that the women in burqas are standing around with their hands on their hips like this, and they have this typical body language of an 18year-old boy. And it is because it has been modified from video games, of course. But we need to start coming up with more cultural specific, gender specific nonverbal behaviors for our therapeutic use, so we are looking forward to that. And this may be non-germane now that some of the other breakthroughs have come through but there are groups out there working on really making mapping of facial expressions to emotions so that we can do a better job of conveying emotions on the avatars and have the right algorithms for doing that. So it doesn't just have to be in real time for catching a facial expression, but then we can generate them as we need them for therapeutic purposes. All right, well, later contact me and I can give you a list of, if you are interested in this area, there are a number of research groups and societies and companies that are doing work in this area. It is really a rapidly emerging area which I think is going to be catalyzed by some of the new breakthroughs in technology that are occurring here. And that is it. That is how they get in touch with me. [applause] >>: I have a question. In the range of applications that you worked with if there was one type of sensing or analysis that you don't have access that you would really like, some part of the body or some particular signal that you would really want to get can you talk about what that would be? >> Walter Greenleaf: Sure. From a physiological viewpoint, I need heart rate. From heart rate I can extract respiratory rhythm. From heart rate and respiratory rhythm I can get sympathetic nervous system tone. From sympathetic nervous system tone I can get attention and I can get anxiety. And if I am going to be doing any telemedicine aspects of this I need that. So I know there is work been done at Media Lab and other places of extracting noninvasively heart rate, but that is something that is very important for all aspects of what I am doing. The other thing that is really important, I have talked about but I don't think it's out there yet, is being able to get facial expressions mapped to avatars. I need a virtual world with train stations and cafés and alleyways and houses that I can have avatars in, but I need those avatars to have proper body language and I need to have proper facial expressions. So it is the nonverbal aspects of human communication that needs to be put on to avatars. What is going on in Second Life right now where people walk like this and their faces like that, just won't do it for therapy. So you asked me for one thing, I gave you two. >>: What about GSR which is important for [inaudible]. >> Walter Greenleaf: You know some people will disagree with this but I am not a big fan of GSR. I think it is, to me it is sort of a noisy signal and it is, some people love it and use it in their research. I always preferred more pulse amplitude and other signals as opposed to GSR. But I wouldn't say no if you could bundle it with some of the other signals. >>: [inaudible] whereas respiration and heart rate don't need body contacts. >> Walter Greenleaf: That's right. I would like to avoid skin contact because clinically I have to do a lot more care with that too, in placing the electrodes and also some people have thicker skin and it is a relative change. >>: Yeah, and it's another signal. >> Walter Greenleaf: Yes it's another signal. Well look, if I get them, I need heart rate; I need respiration; I need GSR; I need EEG; I need eye gaze; there is a lot I need but Jerron only gave me one. But the more psycho physiological signals you can give me, the better. You know, I would love to have like a ballistic pneumograph that somebody could sit on or something on your mouse that has a photo methisnograph that can get pulse amplitude; that would be really great. But I will take anything right now. Yes? >>: So it is pretty clear that these developments work very well in a [inaudible] do you see any application inside of a hospital environment, instead of at home [inaudible]? >> Walter Greenleaf: Well sure, for acute stages of rehab the virtual environments can be really important for motivation. If someone is recovering from a stroke or a traumatic brain injury, that first few steps when they are still in the hospital are really important and the virtual environments can really help with that. It can also later be part of the continuity of care. If somebody is injured over in Iraq, and they go to Germany, and then they go to Walter Reed and then they go to the local hospital, it is nice to have the clinicians be able to track them, and virtual environments can be used for that. I guess also in a hospital--well right now it is used quite a bit by radiologists and surgeons, but I guess I would use it for maybe, well here is one example, if you are taking your son into the hospital the next day for a surgical procedure, you might want to be able to go to a virtual representation of that hospital, walk through the clinic, get familiar, it's stress inoculation, get familiar with what's going to happen, talk to maybe an avatar clinician who is going to give you a tour, prepare you for what's going on. So it is not exactly in the hospital, but it is part of the hospital experience. I think that would be a really great thing. The other way it can be used in a clinic is for, well all I can pick up right now pretty much is neural rehab and stress inoculation but I am sure there are others, pain management, you know. That is an important area too. You must be thinking of one too? >>: I was just curious. So we had a sample [inaudible] duty doing possible discharge, so education. >> Walter Greenleaf: Education is really important I think, again, for clinicians, it's the behavior of being a clinician that is very important. And if you are just using actors to train the clinicians, it is very expensive and they often don't get enough. Also for psychology, it is hard to supervise what goes on in clinical psychology because unless you are using one-way mirrors with volunteer patients, it is very hard to be there. If we start using virtual environments as a therapeutic environment, then a supervisor can be watching as an anonymous avatar, and again with permission, and can observe and maybe even give coaching, we do this right now for our group therapy where the kids might be interacting with each other, but one kid isn't getting it, or one kid is being unruly, so even though they are in a group, the leader, we call it, can talk to the one kid and say look, you're coming on too strong or you are reading that person the wrong way, so even if they are in a group, they can have a private conversation. So the virtual environments involve I guess better group interactions in a therapeutic manner. And that can be done in a hospital setting too. >> Eric Horvitz: I think that's about all the time we have Walter so thanks again. >> Walter Greenleaf: Yeah, thank you guys. [applause]