Hello and welcome. I'm Amy Cohen, the Associate Director for SMI Advisor and a clinical psychologist. I am pleased that you're joining us for today's SMI Advisor webinar, Digital Approaches to the Psychiatric Care of Older Adults with SMI. SMI Advisor, also known as the Clinical Support System for Serious Mental Illness, is an APA and SAMHSA initiative devoted to helping clinicians implement evidence-based care for those living with serious mental illness. Working with experts from across the SMI clinician community, our interdisciplinary effort has been designed to help you get the answers you need to care for your patients. And now I'd like to introduce you to the faculty for today's webinar. Dr. Ipsit Bahia. Dr. Bahia is Medical Director of the Geriatric Psychiatry Outpatient Services at McLean Hospital and the McLean Institute for Technology and Psychiatry. He serves on the APA Council on Geriatric Psychiatry and the Geriatric Psychiatry Committee of the American Board of Psychiatry and Neurology, APPN. And he previously served on the Board of Directors for the American Association for Geriatric Psychiatry and on the editorial boards of five journals, including the American Journal of Geriatric Psychiatry. Dr. Bahia, thank you for leading today's webinar. Thank you for that kind introduction, Amy. Good day to everybody listening and thank you for joining us. I'm honored to be doing this webinar today. I'm going to be talking about digital approaches to the psychiatric care of older adults. Just in terms of disclosures, because this is a technology talk, it's likely that I will mention some brands or companies that you are aware of. I do not have any financial relationships or other conflicts with any of the companies that I will mention. So we will be presenting some research findings that are funded by a myriad of funding sources listed on this slide. So to begin with, let's think about a fundamental truth, that our digital lives and our mobile device interactions are now fundamentally part of our everyday lives. So rhetorically, should they be part of psychotherapy and mental health care for older adults with SMI? Given the title of my talk, of course, that's a rhetorical question. Why older adults? In general, there is this assumption that older adults may not be the most heavy users of technology. This has been studied and the assumption may be partly true. If you see, for example, on the graph on the left, this is from a Pew Research survey from 2013, that with just use of internet and broadband as your basic marker of how much technology is used, that even in the group of older adults above 65, use tends to drop off as people get older. It is worth noting, however, that these are 2013 figures. So the cohort of 65 to 69 is now in the 70 to 74 group and everyone's moved one bar to the right. So even among older adults, use of technology has been increasing. The more interesting finding, though, is the figure on the right. When you compare among just the users how frequently they use technology, the numbers in the 65 plus cohort are much more comparable to people in, say, the 18 to 29 cohort. So 94% of internet users in the younger age groups go online daily. Among older adults, it's 82%. So there is a drop, to be sure, but you can see that once technology is used by older adults, the rate and the frequency at which they use it is fairly similar to younger adults. More research data from 2016 seem to reflect this increasing growth in tech use. You can see in the figure on the left, for example, that the market share among older adults and smartphones is now up to 68% and growing. Among smartphone ownership, among older adults that are affluent or well-educated, you can see also that even though the numbers are lower than younger people, that in the three years between 2013 and 2016, there has been substantial growth. The reason I bring this up is that most older adults are now starting to use technology and especially mobile technology as part of their everyday lives. As clinicians, it's important for us to keep this in mind because we may find that there are ways in which we can leverage technology that they already have and are using as part of the care that we provide when we do it in a thoughtful manner. This is a fact table. I put this only because this is a survey from an adult partial program based in Boston that looked at how different age groups use apps. For one, I'm just heartened as a geriatric psychiatry that they kept track of older adults and how they use technology. You can see that the arrows point to an increasing rate of technology use in the over 60 groups. They studied a range of apps, texting apps, phone or video apps, email, social media, calendar apps, and entertainment apps. We won't go through all of the numbers, but to summarize it, there is growing frequency of app use across the application spectrum and a range of indications. People are owning cell phones and they're using cell phones for a wide range of applications with growing frequency even in the older demographic. You may ask, well, how does the presence of a cell phone lead to mental health diagnosis? This table is intended to demonstrate exactly how this works. We don't expect a phone to actually make a diagnosis of depression, but all smartphones at this point have a range of very sophisticated sensors in them. You can see, for example, that it can pick up movement based on a gyroscope or an accelerometer. A phone can detect location based on the GPS technology built in. And of course, it can detect voice by the use of a microphone. These phones also have high computing power that lets them then process these data. So for example, movement data, which in and of themselves are agnostic, can be processed to determine whether someone is walking or running or standing inert. GPS data may reflect when someone's at home versus when they're working. Phone data, when processed, can tell you how many phone calls a person made. An additional layer of computing and processing can now start to tease out markers like psychomotor activity when you measure it over time. GPS data over time can be a marker of heronic activity once you start to map whether someone was at home or at work or whether they went to a mall or a cafe or a restaurant. Once you aggregate number of calls and track them over time, you can pick up patterns of social avoidance. Now these should start to look familiar because there are three of the criteria for diagnosing major depression. We believe, though, that that last part where you aggregate data from cell phone derived behavior markers and make a diagnosis, that's where the clinician come in. So in an ideal world, the phone would do a lot of the background work. It would provide long-term aggregated data that would help the clinician make diagnoses. This process has a name. We call it digital phenotyping. To give you a clinical example of how this may work, let's consider how we go about our daily clinical practices. This is a simple truth that in the end, even the most skilled clinician is really only as good as the data he or she has to work with. And I will tell you the curious case of Mr. G. Mr. G was an 81-year-old man referred to me for a specialist geriatric psychiatry consultation. He was actually referred to me by his treating psychiatrist. He already had one with a primary complaint of, quote, unquote, brain fog. I did what most geriatric psychiatrists would do in that situation, and I started reviewing his medication list. Mr. G happened to be taking 14 different medications, six of which were sedative. So of course, I said, voila, if you're on six sedative medications, you should have brain fog. So we came up with a plan to actually deprescribe. The approach we take towards that is we tend to go with the least necessary medications first and we aim to get people off of those because that helps to build some rapport and confidence in the process, and it's the least anxiety-provoking. After that, we focus on the least safe medications, and finally, we look at the most essential medications and whether there's any room for changes. So in the case of Mr. G, I first began a taper off of trazodone. He was able to tolerate this without event. After this, we tapered him off of mirtazapine, and he was able to tolerate this too. Medication three on the list was temazepam. Now things got interesting because he had been on this medication for 33 years, loved it, and was very reluctant to get off of it. So the compromise we came up with was that we would deprescribe in 25% decrements. So get him off a quarter of his starting dose at a time. He agreed. So we made the first reduction. He came back to see me a week later, and he tells me that, Doc, that was a terrible idea. I haven't slept a wink. I need my temazepam back. His wife, who happened to be also in all the sessions with him, told me quite the opposite. She said that I was there. He slept fine. So I found myself in this clinical dilemma of whom to trust and which data to go by. The solution was we elected to recommend that he purchase a wearable fitness monitor, one that also tracks sleep. And I said, why don't we avoid the discussion of who's right and who's wrong and let the data from this device give us some sense of how you're sleeping? Mr. G agreed, and he was someone that was relatively tech savvy. He bought a wearable, and he started tracking his sleep. He comes back a week later, and I looked at his data on his phone. Turns out he had slept for about six to seven hours relatively uninterrupted over the course of the night. In other words, his sleep was fine. I happened to swipe the screen, though, and I saw this. This is an actual screenshot of his step count reproduced with his permission. You can see that Mr. G walked an average of 839 steps a day. Much of this number was actually driven by Saturday the 2nd when his wife made him go for a walk on the beach. Without this, his average step count a day was less than 400. I pressed him on what this meant, and that's when it was revealed to me that Mr. G was actually not getting out of bed at all. This was a fact that he had chosen not to divulge to me because he was concerned about what kind of impression he might make. Knowing this, we changed our treatment course. I essentially said that you appear to have neurovegetative depression. He did agree, and he conceded that what I had seen was essentially the best he was at any point in the week because he wanted it that way. Mr. G happened to have been a diplomat in a past life, so presentation mattered to him. Instead of focusing on deprescribing, we went up on his primary antidepressant dose, and he responded. Then, we actually used his ability to visualize his step count as a marker of his clinical progress. Over about eight weeks, he went up to 4,000 steps a day, and it was a pretty significant improvement. He enjoyed seeing his own improvement in a way that was simply not possible otherwise. We published this finding and this case in the American Journal of Psychiatry in 2016. Since then, we have begun to use wearables as part of the standard of care. Here's a more recent case example. This is one of a 71-year-old female with major depression who also had comorbid anxiety. I had the family start tracking her sleep. You can see that at this one point, she was sleeping an average of six hours, 55 minutes a night. Then, I get a call from the family saying that her anxiety appears to be worsened. When they send me her wearable data, it seemed to confirm the fact that she was sleeping almost an average of one hour a night less. In this patient's case, we knew that her sleep patterns were a marker of her level of anxiety. At this point in time, I did two things. I made a medication adjustment, and I requested that instead of sending me pictures of a laptop screen, they send me snapshots or screenshots of her phone, which are much easier to digest. You can see in this latter figure that in the week after, she's now sleeping an average of six hours, 21 minutes, so about half hour more. My medication change appeared to work. However, the subsequent week, her anxiety symptoms relapsed, and she was back to sleeping at just about six hours a night. We made a second medication change. This did not appear to be effective, and now she was sleeping for less than five hours a night. Effectively, we went through a month of communication with me using objective data from her technology to support my treatment decision making. The family and the patient would verify this. I did not actually see the patient in clinic at any point in time. She lived about an hour and a half away, and it was not convenient for her to come and make an in-person visit, but we agreed that we were comfortable. They were comfortable by knowing they had access to me. I was comfortable to have objective data to guide my decision making, so I was just not limited to what they were telling me. Based on this, we elected at that point to hospitalize the patient. It's likely that we made this hospitalization happen three to four weeks earlier than we might have otherwise, and we potentially saved the patient almost a month of suffering. These are commercial wearable devices. The data are not of research grade, but we have found that even just using something as simple as this can really augment treatment decision making in a very meaningful way and give a dimension of objectivity that was simply not possible previously. Unfortunately, though, collecting data to guide how and when to use technology for decision making is limited by the fact that, especially with serious mental illness, older adults are often excluded from studies. An example is this one study that looked at pillboxes and digital or analog reminders that help people keep their medications organized. This was a neat study, actually, that compared four different approaches, using a regular pillbox, using a cap with a digital timer, or using a pill bottle with a strip that you could tear off to keep tabs of when medications were taken. The data seemed to suggest that there is improvement using reminders, and yet the problem here was that in this study, people who are 65 and older were actually excluded. So this remains a common problem with the literature at large, that older adults with serious mental illness, especially around studies of technology, are not part of bigger cohorts that cap their age limits at 65. As a geriatric psychiatrist, this often means that we're on our own, and you can refer to data on younger adults, but as I will demonstrate, patterns of behavior, especially around technology, are not the same. This is something to keep in mind for clinicians, that we're often extrapolating our findings and our guidelines from studies in younger people that we adopt, even though we can never be certain if this is fully accurate. Nonetheless, the converse also applies, that technology studies are not restricted to younger adults exclusively either. This was a study that assessed whether using an app called The Focus, which was developed for augmenting care in schizophrenia, and especially medication use, could improve treatment outcomes. So this was a study with a mean age of 49, now these are not older adults, but these are not exclusively adolescents or very young adults either. And the study found that using an app to augment regular care appeared to lead to improvements in depressive symptoms, it appeared to lead to some improvement in psychosis, and it actually seemed to impact antidepressant medication adherence as well. So even though there are few data on older adults specifically, increasingly we're finding that studies are including people in their 40s and 50s. And this is important, because these are the people that will become our older cohorts in the next decade or so, and it will ostensibly be a much more technology savvy group. So what are the types of data we think about when we talk about data collection and database care? Active data and passive data. Active data are data that are input by the user, usually at predetermined levels. These may be surveys, they may be questionnaires, or it may be regular sampling. So for example, if you have a wearable or a mobile with the capacity to detect heart rate, it may be that actively you have to make sure that you have the device connected so it can capture a heart rate at a specified time. Passive data, on the other hand, are data that are collected without active input by a user. Now these may involve sampling, or they may be continuous. An example would be, say, tracking the number of phone calls someone makes. This is something that all our phones are doing all the time anyway. And as I alluded to earlier in my talk, by tracking the number of calls over time, we may be able to map out periods of social avoidance, or periods of excessive phone use as a marker potentially of mania. This can involve mapping of everyday activities or even sleep. In older adults, we have learned that passive data collection is by far the more high-yield approach. Most of the ongoing research within our group and at other groups nationwide and worldwide are really focused on coming up with ways that passive data can be collected and used in a manner that protects people's privacy and are collected with transparency, but also in a way that clinicians and patients can both access to improve their mental health care. Here's an example of how that might work. This was a study that was published in the American Journal of Geriatric Psychiatry in March. It looked at 1,424 older adults that were previously enrolled in the Health and Retirement Study, or the HRS. This study remains the only one of its kind to actually compare whether the use of four different communication modalities of video chat, email, social media, and text messaging could predict depressive symptoms at two-year follow-up. What the authors found was that of all of these platforms, only the regular use of video chat was associated with a downstream improvement in mood. This was fascinating because we often talk, especially in aging, about how communication technology can help with social connectivity and reduction in loneliness. This study indicated that this is only partly true and that the communication modalities that actually improve mood may be the ones that most closely approximate in-person interaction. So the good news there is that technology can't really replace face-to-face. So the more nihilistically inclined among us should take heart that we will not in fact all be isolated and be reliant on technology. That face-to-face communication appears to be essential. But there is also a piece here that can guide treatment development for older adults, especially interventions around loneliness, that in order to develop tech-based solutions, we need to be more engaged rather than less engaged with the technology serving as a tool and not a replacement for face-to-face or in-person interaction. We also think technologies can help eliminate diagnostic inefficiencies among older adults, especially with SMI. We know, for example, that many studies of depression and aging are limited by the fact that patterns of depression in older adults don't necessarily look the same as in younger adults. This has meant that everything from epidemiological studies that estimate prevalence rates to studies that look at subsyndromal depression and interventions for subsyndromal depression, defined as the presence of depressive symptoms not actually meeting DSM criteria, are all impacted by this. We know, for example, that older adults are less prone to reporting just depressed mood and more likely to have neurovegetative symptoms. What prevalent technologies are starting to do is allow us to measure each of the domains of depression with more accuracy. In this slide, for example, which is based on a publication currently in press at the Harvard Review of Psychiatry, we outline how every single one of the DSM domains can actually be captured by a sensor. We know that there's a few different ways of capturing just depressed mood using ecological momentary assessment, which is regular surveys that capture a mood state at a specific moment in time. There are early data to suggest that voice sensing may be able to achieve this as well. Natural language processing and so on. There are ways of measuring anhedonia, sleep changes, appetite changes. All of the criteria except suicidality, we still are actually quite poor at predicting suicidality based on sensor data. We do know, for example, though, that the best technologies are the most developed technologies are the ones that capture motion. And so any domain of depression that is reliant on measuring changes in motion may be the most reliably diagnosed using digital data. Apathy, for example, or psychomotor retardation or low energy levels. So this is pointing at potentially a new way of thinking about and diagnosing depression in the future, where we move towards a more domain based approach rather than syndrome based approach. But in older adults, I think this is starting to be ready for use. Even now, it will not replace the DSM or the diagnostic ICD approach anytime soon, but we're starting to get there. And a roadmap towards how to use technology, especially for serious mental illness, may look something like this, that you have a patient that has their devices. We look at a combination of existing passive sensors from these devices or other technologies like radio waves or proximity sensors, combine these with active input and surveys to create a digital database that will augment the clinical assessment. And that might actually be something that clinicians get better at incorporating into care to provide treatment and diagnosis. How might this future approach look? So this is a case that I call digital phenotyping in practice. This is the case of an 85 year old female with dementia who resides at an assisted living facility, but also has a comorbid diagnosis where we can't decide whether her behaviors are part of major neurocognitive disorder or whether they come from major depressive disorder. This patient also had a pattern of agitation and pacing. Several people listening to this may be familiar with the clinical conundrum of this sort, that it's really hard to get objective data, especially in people that live in assisted living facilities where we are reliant on staff. This was one of my patients, for instance, and I would see her once every two weeks, often on site. And when I asked how she was doing, more often than not, their response would be summed up in two words. She's fine or she's OK. And then one day this patient was, I was told, hospitalized because she happened to punch a peer in the face over breakfast. And this got me completely off guard because if she had been fine or OK, what happened? Was it truly a random behavioral act that stemmed from her major neurocognitive disorder? Or had there actually been some primary access one comorbid pathology that no one had picked up on? So I started thinking about what we had missed and whether technology or continuous data may actually have helped us predict this episode before it happened. So we went looking around the Boston area, which happens to be a major technology hub, to see if there was something we could find, even if it was prototype. And we came across a technology at MIT called the Emerald. I'm about to play a little video clip that has some sound attached. And I play it because the video clip illustrates the principle behind the technology nicely. So here goes Emerald tracks the 3D motion of a person from the radio signals reflected off their body. Emerald works even if the person is in a different room than the device. Emerald sits in the background and alert. The reason I played that clip was because it shows you how this works, that the device, which is actually just an innocuous black box seen in this picture here, emits a radio signal that bounces off the person. These radio signals are about one hundredth the intensity of regular Wi-Fi. So they're very safe. And like Wi-Fi, they can actually penetrate walls. So you need just one device to cover an entire living area. Why did we go looking for technology like this instead of just relying on mobile phones or wearables? And the answer is that especially in people that have cognitive impairment, which includes a majority of older adults with SMI, mobile and wearable data aren't necessarily reliable. They need to be carried. They need to be worn. They're not always intuitive to operate in this technology. We found something that sits on the wall is plugged into the power main, as you can see. And the person needs to neither touch nor interact with the technology in any way. So even though this is prototype technology and not commercial, we saw great value in using this and then validating it because we think this is how the field is going to move in the future. This video clip shows one of our recent volunteers. You can see he's walking along a spiral on the floor. And he's represented by the red dot in the bottom left corner of the screen. And you can see that the red dot actually reflects his movements very closely. Also of interest is the fact that the device is actually not in the room. The arrow points to the wall that the device is sitting behind. So it's able to track his movements this closely, even though the signals are being transmitted through the wall. So what we did was we installed this technology in the room of Mrs. E. And we've created a second video clip to capture to reflect the data we captured. I will walk you through this. We made a floor plan of Mrs. E's room. And you can see here that she is the red dot. We will start our playback of her data at 146 p.m. on. In May 2017, you can see that she is now entering her room, walking towards her chair area. This is in real time. So it's pretty cool that we can see everything she was doing in her room without needing a camera. And because you wouldn't know who she is, it's the identified at source. When we speed up playback. We can monitor her data far more closely. So she's out of her room at the moment and we calibrated it. So we only capture her own living area rather than the entire facility. This was to protect the privacy of people that live nearby. Now she's back when she's still she disappears. So we only see Mrs. E when she is in active motion. I should add that the device can track four people at a time. You can see again, she's back. She sits on her bed and now she's moved out. But to continue, the device can track four people at a time. And then we did use an accelerometer to capture Mrs. E's gate signature. Once we have this, we can distinguish her from other people that may be walking around her room, like staff, for example, or family members. If you've been watching the clock, you may have noticed that we started at about 146 and now it's 15 minutes later. And you can see that Mrs. E has moved in and out of her room almost eight times in the course of these 15 minutes. This is classic pacing behavior. We still felt that this was too much data and we can't possibly have anyone physically overseeing this in real time. So we developed a metric called the motion episode. Each of these squiggles represents one motion episode, which is defined as a movement uninterrupted of six meters or more in a single line. So every motion episode is one time when the patient moves. If someone were to walk for about 15 meters but stopped in the middle, it would be two motion episodes. And we colored it because that allows us to track individual motion episodes over time. It also allows us to map them. And once we map them, we can really tease out important patterns. We can, for example, tell a low activity day from a high activity day immediately. We can also map out movement from someone that's Mrs. E from someone that's not Mrs. E. So you can see, for example, that there are differences in where in her room Mrs. E goes versus where other people go. This part of her room, for instance, this is where Mrs. E's television set happens to be. And you can see that she operates her own TV, but others don't. You can also see that she seems to get in and out of bed from one specific direction. Now, for people with severe mental illness that may be on antipsychotics or mood stabilizers or other medications that predispose to fall risk, this becomes important because we know that this is the area that needs to be kept clear to minimize the risk of tripping. The same is true for this batch. This is where Mrs. E's chair happens to be. And we know that she gets in and out of her chair, but other people don't. So we use these data, for example, to keep this entire segment of her room clear of all obstructions, like even wires or making sure that any clothes on the floor were picked up to minimize fall risk and eliminate what is effectively the largest risk of medications for serious mental illness in older adults, that they may cause dizziness and increase the risk of falls. It's also important when you look at the data from other people that are not Mrs. E. For example, you can see here that staff members or others were going to this one part of her room and Mrs. E was never going there. This happens to be the closet, but her medications are stored. So, for example, if she was taking medication she was not supposed to, we would have known. We also found that staff members were spending a lot of time on this one specific area of her room at the foot of her bed. We believe that this is because that's where they would stand when they gave her her medications or when they interacted with her, and she would look at them from her bed. It was not a factor for this particular patient, but you can see how if someone had chronic back pain or arthritis, looking at someone standing at the foot of the bed would pose much more pressure on the back or the neck and exacerbate pain, which we know can exacerbate both depression and agitation. So these data can actually help us make subtle behavior changes that lead to more refined and thoughtful care. And when we look beyond just regular practice, let's consider for a second how the DSM-5 and the ICD-10 quantify or give us criteria for diagnosing behavior disturbances in the context of dementia. The topic of today's talk is serious mental illness, but as a geriatric psychiatrist, it is important for us to acknowledge that essentially when we treat older adults, cognitive impairment is always in the background. So it really was not possible for me to focus on serious mental illness without at least bringing the complexity of real-world care and in the context of dementia into the picture. Now, for people that may have dementia, both ICD-10 and DSM-5 require us to make a diagnosis of the dementing or cognitive disorder. They allow us to qualify what the etiology is, but all behavioral disturbances are essentially grouped under one qualifier. Now, we use data from this device and other patients beyond Mrs. E. I will walk you through what this figure represents. So these are a series of concentric circles, each of which represents a 24-hour period. The innermost circle is July 1st, 2017, and the outermost circle is November 1st, 2017. We also use artificial intelligence-based algorithms to separate movement data into time spent asleep, time spent moving about in the bed, time spent sitting in the chair, time spent in the bathroom, and time when a person was out of range. So such is the power of this technology that effectively this one figure reflects all of the movement this person made in their room over a period of four months. So it's really an extraordinary amount of data. This person happened to be a 75-year-old male with vascular dementia, hypersomnia, and severe anxiety. Oh, I apologize for that. You can see here that this person, as represented by the gray, tended to go to bed at about 8 p.m. fairly consistently, remained in bed for 12 hours until 8 a.m., then went out for breakfast. We like to call this the breakfast slice. But right after breakfast, came back into their room, tended to again sleep through about noon. Then there's a little lunchtime slice, and then appeared to go back to sleep again until they had this dinner slice when they spent some time in the environment. You may also see a thinning of the gray as we get from the inside to the outside, representing a series of medication changes targeted at reducing this hypersomnia. So the device allows us to potentially track medication changes as well. But now compare this behavior pattern with a different patient. This is now an 85-year-old female with Alzheimer's disease, pacing, significant circadian disruption, and elevated mood. You can see that the gray is in a wheel, showing us that she was having periods where she was spending all day in bed with like some of these white circles in the middle, which were periods when she was not sleeping at all and spending all her time outside. So it really lets you profile a person's behavior over time in a way objectively that was just not possible to do before. Compare this person with a third person. This is an 86-year-old female with Parkinson's disease and Parkinson's-related dementia and severe psychomotor retardation with comorbid depression. She has a very nice sleep-wake cycle. But as you can see from the amount of green in her day that she spends an inadvertent amount of time just sitting in her chair in her room, not doing much. If you think about it, these are three very distinct behavior profiles. We would treat all three of these people very differently. And yet, under our current diagnostic systems, all three have the exact same diagnosis. I present these data to illustrate how using technology in a thoughtful, targeted manner is actually going to allow us a level of precision, not just in our diagnosis and treatment, but in our entire nosology of mental health in a way that we simply cannot do otherwise. As a geriatric psychiatrist, I consider this especially important because our current diagnostic systems don't capture the variances we see in late life. Even more interestingly, we struggle with the fact that it's hard to find ways to derive behavioral markers from data like this. Until we realized that for over 50 years, mouse researchers have been using exactly this. Capturing the way a mouse moves in an open field as a marker, not just of diagnosis, but really this is the approach that is the bedrock of all drug development in psychiatry. In some ways, our data tantamount to human open field testing. So once we can start looking at human behavioral data in a way that we have looked at mouse behavioral data, our fundamental models of behavior that guide basic drug development will change. Our team actually wrote a hot topic paper on this recently in the journal Neuropsychopharmacology, advocating that motion mapping in humans may be the next great biomarker in psychiatry. In aging, but actually across all serious mental illnesses across all age groups. So this is an exciting time to be working in this space. The technology is also revolutionizing interventions, and there is a small body of literature on interventions specifically for older adults, especially for SMI. This was a study published earlier this year that compared CBT to go, which was a CBT app for schizophrenia. And they compared this with just self-monitoring based on psychoeducation and then mobile assisted documentation to just treatment as usual. And they found that there was a small improvement in psychopathology in the two mobile groups at 12 weeks, but modestly a greater improvement in bipolar disorder compared to schizophrenia. And most importantly, community function appeared to improve. The reason we really love this study is because it points out the possibility that engaging in technologies in a supervised manner as part of care, even when it does not improve psychopathology may actually improve functioning. And this is a very intriguing finding that will be explored more closely. In our group, we've started using technologies just to administer existing treatments in a more creative manner. So, for example, we have a group at McLean Hospital now that used to be a mindfulness group that used basic art therapy, and we introduced simple apps and piloted whether app augmented photo therapy that also included digital photo editing can be used successfully for anxiety and depression. So this was group based care where people underwent a mindfulness exercise. They use digital photographs and then photo editing as expressive therapy. And we found that even the small early work in a group of 11 patients, that there were improvements in functioning when we use the Rosenberg self-esteem scale and the day to day experiences scale. These were people whose psychopathology was relatively stable. So we did not use things like the PHQ-9. But people really loved the fact that they could verbalize internal mood states using simple, easily accessible technology in a way that just was not possible before. I especially love the first figure on the left where someone used the picture to illustrate how depression can sometimes feel like being trapped at the bottom of a black hole. The picture happens to be of a since repurposed elevator shaft. We're also starting to look at whether we can use virtual reality in older adults. So these are new technologies that actually come to us from the gaming world. There has been a sharp growth in research on virtual reality as the graph shows that from 2008 to 2018, the number of papers on virtual reality and aging have really exploded. Most of the literature thus far has looked at using VR for cognitive testing and cognitive training. Very small literature looking at VR as treatment for serious mental illness or even things like PTSD. But we predict that as VR technology becomes easier and more intuitive to use, this will be the next great frontier in creating very personalizable technologies for intervention. So to summarize, if we think about how care works right now, there is diagnosis and treatment as two sides of the same coin. Currently, the standard of care for diagnostics is evaluation, lab tests, imaging and testing. And for treatment, it's psychopharmacology, psychotherapy with follow-up care and monitoring. We're now starting to really incorporate technologies to augment diagnosis. This can be off the shelf. It can be commercial apps. It can be research tools used for practice. It could be prototype technology like the Emerald or web-based tools for testing, many of which are easily available. And they all augment the process of diagnosis. Similarly, with treatment, there are apps for therapy that augment care, apps that support care like med reminders, apps that improve access to care like telemedicine, which is so universal now that we don't even think of it as technology. We actually think of it as standard of care. But then tablet devices or new sensor pills that can actually track adherence or things like ecological momentary assessment or surveys, all of which augment treatment. We believe that this can lead to a model where the divide between diagnosis and treatment will disappear. And we will start thinking of this instead of diagnosis and treatment as assessment and care that work in a constant feedback loop, allowing us to make changes in a more rapidly responsive manner and more personalized and more catered to the needs of the patients. So I think it's safe to say that psychiatry is on the verge of major change. We have to be very cautious around privacy, transparency. So this is not something we should adopt blindly, but we should be optimistic and we should be attuned to how we can use these tools to make care better. I did want to take a quick second to acknowledge our research teams at McLean and Harvard Medical School, but also at UC San Diego, where I worked previously and our collaborators at MIT that are helping us develop this work. I'm happy to be contacted by anyone listening. I do speak on Twitter about our work and more general about the space. And I thank you for all of your attention.

digital approaches

psychiatric care

older adults

serious mental illness

technology

mobile technology

digital phenotyping

wearable devices