Session 2: NIMH 75th Anniversary Event 3

Transcript

SERENA CHU: Welcome back, everyone. If you could take your seats, we're -- I'm going to go ahead and get started. Good afternoon. My name is Serena Chu, and I am a scientific review officer in the Division of Extramural Activities. And it is my great pleasure to introduce the speakers to you today.

We -- joining our panel will be Dr. Silvia Lopez-Guzman, who is chief of the unit on Computational Decision Neuroscience at the National Institute of Mental Health. Dr. Guzman received her M.D. from Pontificia Universidad Javeriana in Bogotá, Colombia, followed by her Ph.D. in Neuroscience at NYU Center for Neuroscience.

She also holds a joint appointment with the National Institute on Drug Abuse, where she is part of the Translational Addiction Medicine branch. Her lab studies the computational and neurobiological basis of decision making and how this process is altered in depression, anxiety, addiction, and pain.

Also joining the panel will be Dr. Alexandra Rodman, who is an assistant professor in the Department of Psychology at Northeastern University, where she directs the Social Development and Well-Being Lab. Dr. Rodman received her Ph.D. in Clinical Psychology from Harvard University and completed her postdoctoral fellowship in the Stress and Development Lab in the Department of Psychology at Harvard University.

She completed her clinical internship at the Boston VA Hospital, rotating through the general mental health clinic and the residential substance abuse treatment program. Dr. Rodman's work centers on the social interaction of teens, examining how social experiences interact with ongoing cognitive and brain development.

Also joining us will be Dr. Ashley Hagaman, who is an assistant professor in the Department of Social and Behavioral Sciences at the Yale School of Public Health. Dr. Hagaman earned her M.P.H. from Emory University, Rowland School of Public Health, and her Ph.D. in Medical Anthropology from Arizona State University.

She is also a qualitative methodology -- methodologist with the Center for Methods and Implementation and Prevention Science and holds a secondary appointment in the Department of Anthropology. Her work focuses on systems-integrated interventions in a global context to address suicide, mental health, and the life course impacts of interpersonal relationships on health and psychosocial well-being.

Also joining us will be Dr. Jane Zhu, who is an assistant associate professor in the Division of General Internal Medicine at Oregon Health and Science University. Dr. Zhu received her M.D. from Harvard Medical School and a master's degree in public policy from Harvard's Kennedy School of Government.

She is a core faculty at the Center for Health Systems Effectiveness and an adjunct senior fellow at the Leonard Davis Institute of Health Economics at the University of Pennsylvania. Dr. Zhu's research centers on healthcare access and quality, particularly for mental and behavioral health services, as well as the effects of provider incentives and organization on healthcare delivery.

So, please help me in welcoming Dr. Silvia Lopez-Guzman to the stage.

SILVIA LOPEZ-GUZMAN: Thank you so much for that introduction. And thank you to the organizers of this event for this invitation. And happy 75th to the NIMH.

So, today I've chosen to like move a little bit away from my comfort zone. I'm barely going to touch on my science, and I'm just going to tell you about what makes me, you know, kind of optimistic about our progress towards delivering personalized care in psychiatry.

Let's see if I can handle this double -- no. I guess this. There we go. All right. And so, I'm going to do my best to summarize the evolution of the field's ideas about individualized psychiatry or personalized psychiatry and the challenges that we've come to face.

And then I'm going to pivot to discussing some promising sources of data, which, combined with the novel data analytical methods afforded by machine learning and artificial intelligence, can maybe help us push forward.

And I'm going to mention, just as examples, a couple of promising, individualized interventions that are starting to be used today. And I'm going to end just on some thoughts of where we're at and what we need to do to get these novel ideas from kind of like the workshop to the implementation and clinical science.

And I'm going to just give a disclaimer that, of course, I'm not going to be talking about research that I've done, right? And in some cases, I'm going to be talking about things that I really don't know much about [laughs]. But I am going to pepper in maybe a little bit of the work that we do in my lab.

Okay. So, of course, you know, this is what we deal with, right? We are interested in trying to answer questions about an individual, right? And the questions, clinically, that are driving us are, can I use information from this individual to reach an accurate diagnosis, to predict and identify how severe the disorder is? What are the risks? What kind of treatment should I select, and how is this person going to do with this intervention?

And so, in the early 2000s, the first efforts of genetic sequencing in individuals with mental health disorders were combined with, you know, the knowledge that there was huge heterogeneity when it came to responses to pharmaceuticals. And so, that sort of, like, propelled the sort of like, field of pharmacogenomics.

But, of course, as you know, while many of these assays currently exist. They haven't really been widely adopted. And this is in part because some of the studies are not very high quality, not very well powered, and there are concerns from providers that what they might end up doing is just kind of perpetuating or amplifying already existing health disparities.

Later, there was another wave, right, of interest into the search for biomarkers. And I would say that in parallel, there were a lot of efforts in trying to get samples from plasma or CSF or other biological samples and try to investigate what are molecules that could be informative of the presence or absence of a disorder or specificity about a particular disorder.

And in parallel, there was a big push for finding biomarkers using, you know, noninvasive methods like neuroimaging. And -- but both of these efforts kind of faced similar constraints of small samples and kind of low reproducibility. And so, clearly, none of these approaches alone have provided kind of implementable tools yet.

And I should say that they have been significant efforts in bridging together genetics and neuroimaging, and we have learned a lot. But we're not yet at the point that any of these, kind of technologies are recommended for clinical implementation routinely. So, why has it been so hard?

Well, at least in part, I think it is due to kind of like immutable features of mental health, the right conditions, namely, that psychiatric disorders suffer from low essentiality and high heterogeneity. And what this means is that there is no single neurobiological mechanism that maps on to the behaviors that characterize a disorder, right? And this is because the brain-behavior relationship is "degenerate," right?

There are multiple behaviors that map onto a single neural mechanism. And then there are multiple neural mechanisms that lead the same behavior, right? And furthermore, these relationships are not invariant to external factors, such as social and environmental context, right?

And then in terms of heterogeneity, what heterogeneity means here is not heterogeneity across people, which we know is a factor, but heterogeneity in the sense that all of these explaining factors of the disorders actually interact in pretty complex and seemingly idiosyncratic ways.

So, where's the solution? Well, at least in part, you know, the solution should be data, right? Because the more factors we measure and we can account for and that we can combine, the more we can account for these complex interrelations. But data is -- alone is not enough. And I and others have argued that we also need to collect data over time, right, to capture, you know, kind of instabilities and factors like development.

We also need to know what context the person was in when we collected the data, right? So, a patient comes in, and we're going to do an fMRI session. We kind of need to know something about what's going on that day. Because there might be factors that we're measuring that have very much to do with what happened to that person coming into the lab.

And then the other thing that I think we really need is theory, right? That will help us kind of organize our ideas about what data to collect and in what way. So, let me now shift gears to, you know, a few examples of promising new data approaches that have become more and more common in the last decade or so.

And, you know, one of the first things that I'm going to talk about -- and I know that this is going to come up in other presentations -- is the fact that now we can use technology to sample, quite with high density, clinical and behavioral data. Here I'm showing you an example of a study following individuals with psychosis and at risk of psychotic disorders and just basically sampling multiple times a day.

There are symptom levels, right? And trying to use machine learning methods to kind of derive different clusters for different types of profiles and make -- and base decision making on them.

You can also combine these sort of longitudinal repeated measures approaches with measures of behavior. And so, here, this is from a study that we performed a while ago in individuals with opioid use disorder that were seeking treatment in an outpatient clinic in New York City.

And what we basically did was repeatedly assess both their symptom levels, but also a number of measures of computational measures of decision making, to try and see if what are -- what combination of these factors is, in fact, predictive of imminent reuse events. And what we found is that, indeed, what you expect is that a combination of these factors play a role in the prediction.

Digital phenotyping is another approach that we can take using smartphone data or online data. This is from Brenda Curtis's lab trying to predict outcomes in addiction on the basis of language in social media posts and showing that, in combination with clinical assessments, the prediction of imminent of relapse was improved.

Now, in addition to what we -- the kind of data that we can get from smartphones, we have wearable devices. Earlier we heard about applications and uses for wearable devices. This is one example. So, this is trying to predict stressed states on the basis of physiological measures from the sensors with high precision.

We too have tried to combine, again, this with behavior and decision making. And what we've been able to show is that measuring things like facial expression and arousal responses and heart rate variability can greatly -- and during decision making, can help us classify or extract when someone is in a particular clinical state, from high craving to low craving.

And then moving away from smartphone and wearables, there is a treasure trove of data in electronic medical records and in all the interactions that users have with providers. So, the United States is not really at the forefront of use of electronic medical records in psychiatry. And there are, you know, real difficulties when it comes to what to do with those data because there's both structured text and unstructured tests.

But this is one example where AI could really be helpful in leveraging that source of data. On the right, I have an example of another way of leveraging the contact between patients and providers. So, this is using machine learning methods during clinical interviews and basically extracting features from, you know, verbal information, nonverbal information, facial expression, and even kinetics, to try to make predictions about group assignments or severity.

And this is really outside of my realm of expertise, but really, there's been an explosion in the field of multi-omics, like the costs of running assays at different levels, from like genetic expression to proteins and molecules, has really gone down. And so, there is a place in which, you know, much like the 2000's wave of pharmacogenomics, these can be used in combination with other sources of data to help -- to aid in precision medicine.

And this is, you know, kind of one of the last things I'll talk about, neuroimaging, right? So, I mentioned in the -- at the beginning that there was a lot of excitement around neuroimaging and finding differences and signatures of brain disorders and in -- that could allow us to separate or differentiate patients from controls.

And what we found -- we found some of those differences. Now, the shift has been going in the -- not in the direction of comparing patients to controls, but kind of figuring out if there are individual signatures of connectivity, right, circuits or biotypes that actually map on to treatment response and to multiple different disorders, right, and clinical manifestation.

And what we want all this data to do is to help us inform intervention. And so, here I'm showing you a couple of examples of individualized interventions. And so, on the left, I have ecological momentary interventions, right? This is like ecological momentary assessments, but the idea is that there are multiple interventions that are delivered through time, right?

This is -- on the left is a study aimed at increasing self-esteem in youth through all these specific tasks that were delivered at different times and showing that these types of interventions can be effective.

And on the right, I'm showing you an example -- an array of different -- and it's really invisible, I apologize -- an array of different commercially available apps that are currently collecting a lot of information on symptoms, sensor-based information, device-based information, self-report.

But those apps, even though they purport to be, you know, for mental health -- and there are some studies that show that using those can be beneficial -- they don't really deliver any type of intervention. And that -- and if they do, they don't tether the information to the state that they can predict that the patient is in, right?

And so, this is the idea behind just-in-time adaptive interventions, which is that we could use, you know, kind of a closed-loop approach to figuring out how to deliver the intervention in the moment that it's needed. Okay. So, now, we have a much more full picture and complex picture of what we might need.

And during the last minute, I'm just going to try to reflect on what it's going to take to go from these exciting new ideas to something that is implementable. And the truth is that, you know, only about, you know, less than half of a percentage of, you know, transdiagnostic models, computational models, prediction models, machine learning models, right, are actually implemented in clinical practice.

And the reason is because we haven't done the validation studies yet. We're not -- they're not sufficiently replicated. So, there's a lot of work to do in that space, and there's multiple challenges ahead. I'm not an expert on this, and I think others will touch on it. But there are -- there's, of course, very complicated ethical issues around the use of AI and machine learning in precision medicine.

Just because we can collect the data; should we collect the data, how are we going to protect the data, is something that is really important. We need to restore kind of this idea of personhood and context and culture and let that inform how we do our studies. And this, you know, has been talked about a lot in the morning about how to involve stakeholders and, you know, kind of like have these looping effects of community-based, informed research.

And finally, we need to figure out ways to lower barriers for clinicians. It's not always easy, even if you want to, to implement these technologies. And so, that -- there's a lot of promise in future efforts to do that. And with that, I'm just going to end. And thank you for your attention.

ALEXANDRA RODMAN: Hello. All right. Thank you so much for having me. It's been thrilling to hear from the rest of you. So, what I'd like to talk to you about today is Coming of Age in a Digital World: Advancing Smartphone Measurement to Predict Adolescent Mental Health.

And so, this is me. Adolescence, as I'm sure you're all well aware and remember, is a time when so much is changing in the brain, physically, psychologically, and of course, socially. And in addition to taking on new roles and experiences, adolescents are notoriously preoccupied with peer approval and social belonging. And every interaction can feel more intense, whether it's the excitement of a school dance or the sting of being turned down by your crush.

So, back to 13-year-old me. This picture is actually from a group photo shoot for my friend's birthday. And this memory is so incredibly strong. You'll see the limited [unintelligible]. That was -- that's real. This memory is incredibly strong. Because I remember feeling like this friend group was locked in, right? Like we had done a professional photo shoot. And we were going to be best friends forever. And that felt really, really good.

But little did I know that some of these ties wouldn't last, and the falling out would be pretty painful and all-consuming. Typical adolescents, right? But why is that? What is it about adolescence that makes social experiences so intense? And on some level, it has to do with how we process social information.

Just like any other kind of information, we don't just absorb social information as a one-to-one, input-output process. Instead, the information we perceive is filtered through our own individualized lenses and is subject to distortion or bias due to prior expectations, motivations, and interpretations.

So, when it comes to having varied social experiences that range from positive to negative, like being invited to a party, feeling left out, the way we process those experiences may take different forms. And positive and negative experiences could be weighted differently. And this weighting matters because the biased information is then integrated into more global impressions of how we view ourselves, others, and the environment, guiding our expectations going forward.

So, in my work, I take a developmental perspective and focus on the age-related differences in social processing behavior and experiences to try to understand adolescents' unique social sensitivity and what happens when things go awry.

The transition from childhood to adolescence, typically occurring around age 12, is often described as a period of social reorientation. Adolescents spend less time with their families and more time with peers. Social belonging becomes increasingly important, where teens are more dependent on each other for support. And meanwhile, social groups are in flux during this time, subjecting adolescents to greater frequency of peer rejection.

This reorientation is accompanied by enhanced intensity of social experiences that can shape emotions and behaviors in meaningful ways. Adolescents are more preoccupied by peer approval and show more intense emotion and stress responses following peer rejection. And while rejection is especially common during this time, peer rejection and a history of it has been linked to depression.

Indeed, most adolescent mental health problems emerge in the aftermath of interpersonal stressors, such as being bullied, a romantic breakup, or peer conflict. And at the same time, adolescence is a period of peak risk for the emergence of mental health problems, particularly depression, anxiety, and suicidal behaviors.

At the start of adolescence, the prevalence of mental health disorders is at about 1 to 2 percent. But by the end, this increases to 25 to 3 percent. In fact, 75 percent of all mental health disorders will emerge during the adolescent years. And this sharp rise in mental illness spans many disorders, with the average age of developing any mental illness falling around 15 years old.

Importantly, onset at this time tends to have a more severe and lasting prognosis into adulthood. So, with all this in mind, clinical and developmental psychologists think of adolescence as a period of vulnerability, when social sensitivity comes at the same time for -- the same time as risk for mental health problems.

And while I've pointed out the clinical risks associated with this period, there's also tremendous opportunity. Adolescence is when clinical interventions stand to make the biggest impact, ultimately shifting trajectories of risk.

This is due in part to widespread changes in brain structure and function, where the developing brain is making and pruning connections based on life experience. And these ongoing changes contribute to the social sensitivity and clinical risk we see during adolescence.

But at the same time, these changes support dramatic growth and learning. And this enhanced neuroplasticity, when the brain is particularly responsive to the environment, provides, again, a window of opportunity where interventions can make the greatest impact, especially when compared to adulthood, when most people first seek treatment for mental health problems.

And given the substantial change that occurs in the domains of social experience, brain development, and risk for mental health problems, my lab's work centers on the interplay of these domains, studying how social experiences interact with ongoing cognitive and brain development to shape adolescent well-being and clinical risks. Sorry.

Our primary questions explore what makes adolescents uniquely sensitive to social experiences, how social factors function as mechanisms of risk for psychopathology, and which factors enhance resilience against mental health problems?

And to answer these questions, we take a multimodal approach, including experimental and observational approaches, using novel behavioral tasks, fMRI neuroimaging, and digital phenotyping via mobile phones.

And while adolescence has always been known as a period of immense change, what that looks like today is rapidly evolving due to the sudden omnipresence of the digital world. Soon enough, teens' real worlds and digital worlds will become fully integrated, if they haven't already. So, it's critical that we begin to study adolescents in new ways that fully grasp the complexity of the adolescent experience.

And so, for today, I'm not going to talk so much about my brain research, though I love it. And of course, we aim to integrate it with our real-world measures. Instead, I'm going to focus on our work using experimental tasks to probe adolescent sensitivity to social experiences, followed by our past and current work examining smartphone use and its direct and indirect associations with stress, mood, and anxiety.

So, first, I'd like to share some work investigating the social motivation and learning processes that may render adolescents uniquely vulnerable to peer stressors. In a study of 102 participants, aged 12 to 23, we use a physical effort paradigm to see how hard adolescents were willing to work to find out if they were liked by their peers.

And in this figure, I show you their expectations of being liked on the X axis and how hard they squeezed to obtain peer feedback on the Y axis. And as you can see, adolescents squeezed harder than adults overall but worked hardest when they thought they'd be more strongly liked or disliked, working the least for expected neutral feedback. Adults, on the other hand, worked harder the more they expected to be liked. And next, we were interested to find out what adolescents and adults did with that feedback once they received it.

In a study of 107 participants, aged 10 to 23, we asked how peer acceptance and rejection impacted self-esteem. On the Y axis is age and -- I'm sorry, on the X axis is age, and on the Y axis is change in self-esteem from before to after the task. We found that after receiving the same amount of positive and negative feedback, teens showed a drop in self-esteem, while adults showed a curious boost in self-esteem.

So, across both studies, adults show a self-protective response, whereas teens valued strong signals of inclusion and exclusion, even though their self-esteem is particularly impacted by it. And next, I became interested in investigating social behaviors as a mechanism of risk.

And in this next study, we focused on phone call behaviors. Here, we took a within-person approach, examining how monthly fluctuations in these behaviors explain the relationship between stressful life events and anxiety symptoms over the course of a year in 30 adolescent girls. We found that when individuals experienced more stress than usual, they also engaged in more phone calls than usual.

But what we found next was surprising. We found that more incoming phone calls in turn related to greater anxiety symptoms the following month. Moreover, calls mediated 40 percent of this stress-anxiety relationship. And this was the kind of study that we love and hate [laughs] that just left me with more questions than I started out with.

I had expected that calls would help mitigate the negative effect of stress on anxiety; reasoning that, calls signaled social support. But our results showed the opposite, which turned out to be a blessing. I wondered, "What were these calls? Was it the stress itself? Who were they with?" And in the end, it really impressed upon me the importance of getting more granular data.

So, we started to scratch the surface of this in a study examining social factors that enhanced resilience. During the COVID-19 pandemic and a longitudinal sample of about 250 children and adults, we found that the relationship between COVID-related stress and depression and anxiety symptoms, or internalizing symptoms, six months later, was positive -- positively related in youth, who decreased or showed no change in how much they socialized over digital platforms during the stay-at-home orders.

But for those who increased socializing over digital platforms, this relationship was effectively erased. And we found that this effect was primarily driven by phone calls compared to other digital means of socializing. And in a recent preliminary study of nearly 30 adolescents followed monthly for up to a year, we not only examined different types of phone use, like communication, social media, gaming, entertainment, but we also examined different metrics of phone use, like screen time, pickups, notifications, all of which had differential associations with positive and negative mood.

And importantly, we did not find that oft-touted relationship between overall phone use and negative mood. We found actually a very small positive effect. And the primary takeaway here -- and I know I'm not going into a ton of detail -- is that -- I would say the primary takeaway is that this is a nuanced picture where different types of phone use likely serve different purposes. And we need to start measuring and analyzing phone use in a way that adequately captures the complexity of the behavior. And some methodological advances in recent years are finally allowing us to do just that.

So, to that end, I'm thrilled to tell you a bit about some upcoming work. We're currently launching our first intensive longitudinal study, co-designed with a teen advisory council, where 80 teens will complete eight monthly visits to gather intensive social and emotional functioning data. They'll come in for a lab visit day, where they'll complete several behavioral and fMRI tasks.

We'll also conduct ecological momentary assessments or brief surveys pinged to their phones and several week-long bursts to ask about emotions, behaviors in real time as they make their way through their day-to-day lives. And finally, we'll continuously and passively collect mobile phone sensor data, including GPS, call and text logs, activity, physiological data from smart watches.

I'm also happy to talk about the ways in which we've designed our recruitment and data collection processes to increase feasibility, access, and representation. I'm also excited to share that this R00 has allowed me to pilot a collaboration with computer scientist Varun Mishra.

Together, we hope to advance smartphone measurement by pairing objective smartphone data with qualitative reports of the nature and function of smartphone use in real time. So, as smartphone time series are coming in, an adaptive machine learning algorithm will consider these patterns and contextual factors to identify anomalous patterns of activity, which have been found to be predictive of negative outcomes.

And this will then trigger an EMA survey to ask participants about this anomalous event; who was it with, what was the purpose, the function, how they felt. And this design helps us move past some of the early barriers to progress by using objective measures instead of self-reported estimates, focusing on within-person effects instead of individual differences, and deep phenotyping by tagging objective smartphone data with this descriptive information, all the while we get these perspective bidirectional relationships at multiple timescales.

And so, we believe this level of granularity will be critical to advancing measurement of smartphone use, that, in turn, is associated with well-being, ultimately predicting when adolescents are most at risk for negative mental health states, leveraging data-driven approaches to enhancing resilience when they need it most. So, thank you, Silvia, for teeing me up. Not for nothing, we'll be submitting this R01 proposal soon. So, wish us luck and stay tuned. Thank you very much. [laughs]

ASHLEY HAGAMAN: Oh, boy, I'm going to start talking about places where there are no smartphones. So, buckle up.

[laughter]

Thank you so much for the opportunity to speak at this event with so many other inspiring and amazing researchers. My name is Ashley Hagaman. I'll be sharing with you today some of the pragmatic innovations that we are bringing into the future of suicide prevention research around the world.

I work with so many amazing Pakistani and Nepali scholars and doctors. And I wish I could introduce you to every single one of them, but I can't. So, here they are, and none of this happens without them.

So, to begin, more than 700,000 people die by suicide every year. Even larger, 16 million people attempt suicide every year. This doesn't account for the families and the friends and the social networks that endure the rippling effects of those acts. Importantly, up to 75 percent of those deaths happen in low- and middle-income countries, or LMICs.

In LMIC settings, 1 out of 10 people that need mental health care, only one will receive it. This disparity is magnitudes larger than the important disparities that we have here in the United States. However, despite the disproportionate burden of suicide in LMIC, our knowledge and interventions are heavily anchored on western theories.

Our amazing librarian at Yale did a search for us, and last week found that, crudely, 11 percent of suicide research is conducted in LMIC. And based on the evidence we do have, we know that risk and recovery profiles differ across culture and context, which is why we see a lot of global variation in suicide rates. So, there's incredible disparity in burden and science.

And the last thing I'll say about suicide, at least more broadly, is that in Southeast Asia, where I work, there are far more female suicide deaths than in any other region of the world. And it's an important outlier and one that we try to address in our research programs.

I want to say a few things about my own lived experiences and motivations. I was trained as an anthropologist. And yes, you're about to hear about a lot of clinical trials from an anthropologist, so also buckle up.

[laughter]

A part of that means that I get to live and learn from families while I do my field work. I have learned so much about maternal mental health, about family, and about care systems from the mothers that I got to live with. I studied suicide. I studied loss and complex notions of agency around mental health, and I got to do it alongside what became family. And I afford a lot of my own career to these amazing women that housed me for years in Nepal.

And as I became a professor and a principal investigator, I also became a mother. And I suffered a lot of personal suicidal loss and challenges within my own social networks. And a part of what makes my job the absolute best is getting to do it alongside these amazing, strong, and seriously smart women, mothers, and scholars.

And so, one of the things that my future and suicide prevention research is defined by is my ability to amplify the amazing commitment, tenacity, and bold work that these women lead. So, I want to start my talk off with dedicating it to them and say how excited I am to be a part of advancing suicide research with them in the coming decades.

And so, my talk is about the future of suicide prevention work. And I hope that you see it in the bold women and youth that are leading it and what it's looking like in some of the corners that we work in Nepal and Pakistan. So, naturally, as someone that does research in places that I do not call my own home, it's hard for me to answer what that future looks like in 25 years.

And so, I asked [laughs] our community advisory boards, our youth advisory boards, my colleagues, and communities what they thought that should look like. And they want their own cultures and explanatory models to become pivotal parts of suicidal theory. They want people to feel like they can disclose their thoughts without harm. They want local experts and local leaders.

And I'm so happy to say that our trials, the ones that NIMH is funding, are doing all of these things. We're creating local leaders. We're creating community-driven interventions using local theories. And I'm so proud that the women that were researchers a decade ago with me are now finishing their Ph.D.’s and are named co-investigators on the trials that we're now running.

So, that, to me, is one of the most important things that the future of our research can do. It can put in positions of power those that have been most neglected from the research. So, we're making it happen. And a part of what makes it happen is grounding all of our co-design and clinical trial work in lived experience advisor reports [laughs].

Thank you, Dr. Bellamy, and everyone else who has really highlighted this point. I think if you get a main theme, this is probably it. And so, they really guide all of our processes and procedures. Our team see that the feature of this work is anchored in culture and context and how to implement. The individual is so important and so is their community and the systems and the policies and the structures that they exist within.

So, for example, in Pakistan, suicide attempt was only decriminalized less than two years ago. So, prior attempts were punishable by imprisonment and fines. So, we cannot build a suicide intervention or ask people about their suicidality without considering and attending to the legacies of these policies. I have lots of other examples here.

But a part of our strength as anthropologists and an interdisciplinary team is that every part of the work that we do considers these elements to optimize our interventions efficacy and likelihood of success. So, one of the ways that we've done this is to develop indigenous theories of suicide, led, of course, by our indigenous scholars. And so, these grounded theories become the foundation of our intervention and implementation work.

And so, I'll give you one quick example, hot off the press -- and like everything is hot off the press because all of this work is in progress -- from work that's led by Gul Saeed and Sidra Mumtaz in Pakistan. And we sought to understand how women think and talk about a life worth living and what transitioned them from suicidal thoughts to attempts.

And so, one part of that theory -- I won't explain the whole thing [laughs]. You can read the paper. It really focuses on how women long for sabr and sukoon, patience and peace. These are states that women describe having strength and control. So, we attend to this in both our measurement of suicidality and management of suicidality.

And there's lots of other examples that are coming out soon from Renu Shakya and Kripa Sigdel from our work in Nepal. I'm going to walk you through the next four years of research trials that my lab is running in Nepal and Pakistan. Importantly, these are only -- these are the only [laughs] suicide prevention trials at the community and primary care levels that have ever happened in either of these countries.

So, they are important reasons why they're pilots. They're developmental, and mostly that they're carefully co-designed and measured. And so, we create packages of suicide prevention programs and design these to be deployed in various settings. We've done it in a large Nepali hospital system. We're currently running pilot clinical trials in the primary healthcare system in rural Nepal and for youth in a community in Nepal and for mothers in Pakistan.

And we designed these complex interventions and their implementation protocols by intervening on the individual at risk, but also intervening on the families and the clinicians and the systems that they're a part of. So, in this way, we're doing what we call hybrid trials, where we study both how effective the intervention is on the individual at risk for suicide and how implementable and potentially scalable and sustainable it is in the real world.

So, we study clinicians and the peers that are delivering the intervention in the systems that they work and live. So, what's in the package? It's been deconstructed and reconstructed from several brief interventions shown to have evidence base for addressing suicide, at least in the west.

Some of these elements are from the Zero Suicide approach in western health systems. Some are from the Mental Health Gap Action Program, which is a program that equips primary care physicians to detect and treat common mental disorders, including suicide. And there are three basic elements.

The first is screening. And we can't think about screening without thinking about the risks and benefits that someone has in disclosing suicide risk. So, we developed systematic, structured screening tools, including both psychometric screeners and strategies for identifying risk at the community level, which is what we call community-informed detection.

And we're validating these instruments and ensuring safety for disclosure. We also leverage evidence-based practice of safety planning or crisis planning, but we've completely reconstructed it to make sense in Nepal and Pakistan. And I'll show you how.

And finally, there's evidence that basic connection and contact addresses important elements of suicidal risk. And we pair this with ongoing support for the person's preferred care-seeking pathways to ensure that they're supported in ways that they want to be over time.

The first trial just finished a couple months ago. This was the original development of the suicide prevention package and the design of an implementation in a Nepali hospital. And it's a really nice example of how we thought about identifying someone at risk and how to support them in the next steps.

This is the emergency department that housed the project. It's generally super chaotic. And you can see this is like a really not chaotic day, by the way [laughs]. But you can see that families surround patients almost at all times. And that's really typical in clinical encounters. It's actually required in the Nepali health system that you have an accompanying family member with you at all times. Because they have to go buy the IVs. They have to go buy the medicines, and they have to bring everything to you.

And so, it's important to understand what screening would actually look like in this context. And our community advisory board was given all of the screening [laughs] tools and strategies that we could find to identify suicide risk. And they went through all of them, and they picked the one that they thought was most appropriate and then we heavily culturally adapted it.

We did this in all of the settings. Here, they chose to ask suicide screen, which was developed by Lisa Horowitz and colleagues. And this is what the clinical workflow of the screening tool looked like. The most important part was training of staff for how to introduce the questions, especially given suicide really isn't ever discussed in biomedical care settings.

And so, our CAB and clinicians work together to develop the script to introduce the questions that were coming, why [laughs] they were being asked these questions, and then what might happen in order to optimize their comfort in disclosing. So, our CAB helped co-develop the workflows and implementation blueprint and train the clinicians and like all -- with them all along the way.

Briefly, some of the outcomes of that initial screening, 14 percent screened positive, which is higher than what we see in most U.S. hospitals. And around 2 percent were at really high risk, meaning they had active thoughts and plans for doing suicide in the immediate future. That's about probably the same as the U.S.

We embedded a validation study into the psychometric analysis of the tool, suggested that it's reliable. Our qualitative work uncovered some really important implementation distinctions from the western settings. I won't go through all of them. But to give you a quick flavor, the ASQ toolkit says that a screen takes less than 20 seconds, which, like, if you role-play it, it can.

But this is really different in Nepal, where folks are far less conditioned to answer serious questions about death with a one-word answer. And so, staff needed time to listen to stories, whether someone was positive or negative. And so, again, we need these really important implementation strategies to make this work.

So, once we know that someone might be at risk or that they've had a recent attempt, safety planning is a really effective way to help them and their support networks know that there's always something they can do to keep them safe.

In the U.S., you can see that these plans are really homeworky. They're quite structured. And they rely on handwriting and literacy. In Nepal and Pakistan, it's common that folks cannot read or write at all. So, we wanted to, again, think of deconstructing and decolonizing these tools to make them implementable and useful.

And the last thing that I'll say about safety planning in general is that we don't really know the mechanisms by which it works. We think it works differently for different people. So, a part of what we're doing -- and we know it might work for an individual. But it also works for families. And it also works for the treating clinician or the provider. But we don't really measure how those work and how effective it is for each of those individuals.

So, that's a part of what we're doing in some of these trials that are coming up. So, in another clinical trial that NIMH just funded a few months ago, we're enhancing the already existing mhGAP package of strategies that I talked about earlier for responding to suicide so that primary care physicians -- they've already been trained in it, but they're not doing it.

So, the whole clinical trial is to figure out how do we get them to do it better with more support. So, what does safety planning look like with some of our illiterate populations? First, we don't call it a safety plan. We call it an ashako diyo, which literally means light and hope, which is really beautiful way, I think, to think about what this instrument can do. But also, it's not quite so awkward, like a safety plan.

Second, we worked with indigenous artists to create stickers that visualized various components of the plan, so it doesn't rely on writing and reading. You can see some of the stickers that are here. The pocket card with those visuals helps both the person that's facilitating the safety planning and the person that's receiving the safety planning sort of understand what's going on and what they're working on together.

And we task shift a lot of the work that physicians were originally tasked with in the original mhGAP protocol to community health workers. So, the original mhGAP protocol says that the primary care physician should follow up with the patient every two weeks. That is completely infeasible in many of the contexts that we work. And so, we get community health workers rather to do those visits at the patient's home themselves.

And so, we'll measure how it worked, both with the high-risk patients that we enroll in our trial and with the clinicians to see how well and effective and with what quality they can implement it.

We also got funds two months ago [laughs] from NIMH to develop a community-based peer package for youth suicide in rural Nepal. And so, the trial has similar elements, but our youth advisory board, who all has lived experience with suicide, and our partner, SOCHAI, which is a women-led social change organization that's been pioneering innovations for public health all over Nepal, are adapting the package.

And they're thinking about ways to make the plan into indigenous jewelry or songs or other materials that don't rely on something that is written. Another key part of the trial is that it's protocolizing helping first families when they have a youth in their household who's at risk and then also figuring out how and when to engage them.

So, this is one example of how NIH is funding and supporting innovative futures and research because two of the named co-investigators on this project are experts by experience. So, they don't have Ph.D.'s or M.D.s, but they have incredible and irreplaceable expertise to make the work succeed.

And the last trial I'll tell you about is our suicide prevention package designed by women for women, delivered by peer mothers in rural Pakistan. It's another hybrid type 2. So, that means we're studying both how effective it is on the suicidal mothers and how well the peers can implement it and make it sustainable within their health system.

It's called the Khushal Pur Ameed Zindagi Trial, which means happy, hopeful life. And our CAB with lived experience picked a different screening tool, the patient safety screen, and developed scripts for patients and the families. There's like a whole family engagement part of the intervention because it's just so necessary and work like this.

And really cool, we designed new tools to measure suicide severity that are anchored in the cognitions, emotions, and moralities that are wrapped up in how individuals think about suicide. So, you have to stay tuned for all those tools and measures and validation papers that are quickly coming.

Just really quickly, the safety planning looks really different. Again, in Pakistan, it's actually anchored in a narrative story approach, where we have a visual story of a woman named Shu Gupta, who's a mom, who, you know, wrestles with her own suicidal thoughts and behaviors. And it's a really effective way of engaging someone in safety planning and sort of this safe but kind of indigenous way of knowing and learning way.

I want to say a few things about how we're advancing methods. We're developing new measures to quantify suicide severity and mechanisms of intervention action. We're developing systematic measurement strategies to assess fidelity and quality that primary care physicians and peers deliver in their evidence-based practices.

We thought a lot about models of supportive supervision to support task shifting suicide prevention work to peers so that they're confident, safe, and cared for. Every single person involved in our project in any way receives free mental health care, hands down.

The last thing I'll say is really about the persistent challenges, I think, especially doing work in low- and middle-income settings, that we face in a funding system that's really full of inequities and limitations.

The first is that the future of innovation depends on recognizing diverse kinds of knowledge and expertise. And NIH is doing this, and we can't be more excited.

We need to design reimbursement structures that are anchored in social justice and attentive to inequity. Rural partners don't have social capital to pay rent or salaries for months on end, waiting reimbursement. Our implementing -- or capital just doesn't exist in rural areas.

One of the starkest inequities and realities that our partners face is really limited indirect rates to keep the lights on and pay for key administrative services. There's an incredible amount of hidden curriculum and grant systems and legal requirements.

And it's -- another thing that NIH is supporting, through one of the capacity-building arms in one of our trials, is to actually create a training system between Yale and our local partners to submit and support the highest quality and most innovative research we can. And would I really be a researcher if I didn't ask for more money?

[laughter]

In reality, we're working in context where, default, there's three or five languages. And we can't just have an English version and an Urdu version. We need multiple versions of all of these things. And we never have the capacity to do that.

So, there's just some of the things that we wanted to highlight to continue to build equity in the future of mental health innovation. So, thank you so much. I'm so grateful to have this opportunity and for our partners that make this work possible.

JANE ZHU: All right. Is lunch hitting everyone? Because if it is, please feel free to stretch and stand up. My name is Jane Zhu. I really appreciate the opportunity to be here today with you all.

So, you know, this morning, we heard a lot of amazing presentations about the forefront of science, new treatment modalities, precision medicine. But I'm a primary care physician. And the sad reality is that I can't even get my sickest patients in to see a psychiatrist with the current availability of medicine that is existing.

So, you know, we've reached an all-time low in uninsurance in the U.S., and yet, everyone in this, you know, room knows that access to coverage, access to -- or availability of these treatment modalities has not translated into access to care. And so, that's what I'm going to be talking about a little today.

So, it turns out that people with the same insurance experience vastly different utilization and mental health outcomes. And so, when we look at mental health utilization rates in Medicaid, for example, the largest single payer of mental health services in the U.S., we find an eightfold difference in the highest versus lowest used areas in the country.

So, I think this figure raises some interesting questions. Why is it, for example, that Duluth, Minnesota, is an outlier in outpatient and emergency visit rates for mental health duress? Or is this an area with high healthcare capacity but relatively low quality of care?

Similarly, what explains relatively low emergency room use and above-average outpatient mental health use in Texarkana, Arkansas? Is this an area that's appropriately shifted its resources to the outpatient setting, or is this an area with relatively mild to moderate mental health conditions, as opposed to severe mental illness?

The answers to these questions are yet unknown. But by studying enrollees who move from one region to another, researchers have actually been able to disentangle some of the factors contributing to this variation. And they've attributed about 60 percent of variation in healthcare use to place-specific factors. So, these are things like provider supply, regional organization, systems of care, as opposed to patient- or individual-specific factors like personal health risks and preferences.

One of the place-specific factors which I study and which I believe to be understudied is this role of state policies and insurance design, particularly through managed healthcare. Managed care is a health insurance approach that is now the dominant form of healthcare delivery and financing in the U.S.

And importantly, managed care organizations seek to control costs and ensure quality of care through various levers that intersect with market and state political and regulatory environments, state procurement processes, and managed care plan structures like profit status.

These functions that managed care organizations perform might include utilization and clinical management, so monitoring covered services and use, including prior authorization, other gatekeeping measures, establishing provider reimbursement rates and payment procedures, and contracting with select groups of providers -- provider networks that deliver care to plan enrollees.

And so, I'm going to be focusing on one of these functions just to illustrate the importance of this to patients. So, when we look at networks of psychiatrists across healthcare markets, we find that these networks typically contain low shares of psychiatrists that are available in a given service area.

On average, across Medicare Advantage, Medicaid managed care, and the ACA individual marketplaces, we find that about half of these networks include fewer than 25 percent of all available psychiatrists in a given service area. Further, psychiatrist networks are consistently narrower than those for primary care physicians and all other specialties.

Let's zoom in further from the network level to the provider level. And we get a sense of how these in-network providers are actually distributing their services. Most dermatologists, as you can see in the top here, are seeing small volumes of Medicaid patients. But it's fairly distributed on that low end.

If you're a primary care physician like myself, you're seeing a pretty even case mix across these insurance types. And if you're a mental health provider, there's almost a bimodal distribution between people who see zero Medicaid patients and people who see almost all Medicaid patients.

So, one important implication here is that not only are we seeing narrow networks of providers to begin with, but even smaller sets of core providers are delivering a disproportionate share of mental health services, particularly in Medicaid.

Let's zoom in further to see what the patients experience. Managed care plans are mandated to provide transparent information about who's included in network for their plans. And if you're, you know, shopping around for health plans. For example, many of you may have used these provider directories to choose your providers to select for plans.

And so, we compared consumer-facing provider directories to administrative claims data in Medicaid managed care. And what we find is quite jarring. More than two thirds of mental health prescribers -- so, psychiatrists and psychiatric mental health nurse practitioners -- and more than half of psychotherapists that plans list as a network, are in fact ghost providers who don't care for these patients at all.

These findings in aggregate have been replicated in other insurance markets beyond Medicaid confirming that patients in reality are facing an ever-narrowing funnel of available providers that they can actually access. There are, of course, important push and pull factors that contribute to the narrowness of these networks.

From the perspective of managed care plans, broader networks on the one hand, provide important patient goals like provider choice and expanded access to providers. Some states may have also any-willing-provider laws, which require insurers to accept any provider in their healthcare plan that wants to, limiting the extent to which they can exert control over their provider networks.

On the other hand, insurers might argue that narrow networks can be used to direct patients to clinicians or facilities that are known to be higher performing or higher value. Narrow networks give insurance plans bargaining power to lower provider prices. And on average, plans with narrower networks perform better financially than those with broader networks.

There's also, of course, important supply side push factors that contribute to low network participation among providers and contribute further to this problem. These factors include, obviously, significant reimbursement disparities, both between mental health services and compared to physical surgical services and across insurance markets and provider types.

Billing delays and other administrative barriers, including those associated with network contracting, credentialing claims processing. And clinical complexity and work environment factors that are often more favorable outside of insurance systems.

Importantly, mental health professionals have also had a longstanding and viable alternative to insurance participation in the form of a robust cash pay market. And in fact, shares of total healthcare encounters, as well as physician-level revenue, have increased in time for private pay amongst psychiatrists compared to, for example, primary care providers.

And the existence of these markets is advantageous in many ways to providers, further disincentivizing providers from participating in insurance. Ultimately, the summation of these factors are mental health networks that are wholly inadequate and exert real effects on patients that have insurance.

So, we see higher out-of-network use and out-of-pocket costs. We see that they have limited access to higher quality providers and team-based care, reduced healthcare utilization overall, and treatment delays and foregone care altogether.

So, the very real impact of this problem highlights the need for future research. And I just wanted to, you know, bring up the fact that these place-specific factors have traditionally been understudied particularly around state policies and managed care design and how they influence how patients navigate the healthcare care system and what this means for their health outcomes.

We need to understand which supply side factors can be leveraged meaningfully by these organizations, by states, to right-size networks with the right services to meet local population needs. And so, for me, a number of really, really simple questions remain, but have yet been unanswered.

The first is the extent to which unmet mental health needs in the U.S. are driven by an aggregate shortage of mental health providers versus -- as I've discussed today -- potential misallocation of providers that exist either through, for example, managed care network design and low insurance participation among providers.

Second, we don't know to what extent these providers are substitutable and what services and provider types should be targeted to which populations. Understanding how we can better allocate treatment resources across the existing workforce, obviously requires a different set of policy responses than expanding the behavioral health workforce in the longer term.

And third, policymakers have thus far prioritized a number of interventions under the premise that a set of modest incentives will induce provider change, whether it's, you know, where they're practicing, whether they're accepting insurance or, you know, who they're taking care of.

But these interventions really need comprehensive and robust evaluation to understand which programs, which policies can consistently improve workforce capacity as it exists.

So, I'd like to end with a brief example of the research opportunities that exist in this sphere. Local -- so, low reimbursement rates in Medicaid, particularly in mental health services, has been widely accepted as a major factor contributing to provider shortages and reluctance to accept insurance.

We found that Medicaid pays on average 80 percent of Medicare rates for psychiatric services with wide variation in payment across states. But given the unique constraints of the mental healthcare system, we're not sure to what extent rate increases can induce changes in behavior.

And there's no empirical evidence that suggests this whether they incentivize new providers to come into Medicaid or they better sustain core providers that already serve these populations. And what magnitude of rate increases are really needed from a policy perspective.

But of course, as we would expect, policy changes move faster than evidence generation can keep up. And so, as of 2024, for example, 26 states have implemented rate changes for behavioral health services as a way to acutely improve access to behavioral health care.

And these rate changes really range. They range from across the board, a whole magnitude increases for all providers to targeted rate increases for select services or provider types.

And this provides a real opportunity to leverage natural experiments in the real world. As those of us who work in Medicaid have heard, "One Medicaid program is one Medicaid program. One MCO, or managed care organization, is one managed care organization." These are really laboratories from which we can learn a lot about policies.

And then against a relative dearth of evidence that currently exists, future research should really prioritize timeliness and dissemination of data that policymakers need to prioritize which policies yield the greatest effects. Or as an alternative, to reallocate resources away from efforts that aren't very effective.

And finally, to conduct this research as a researcher, I'm very grateful that we actually have unprecedented access to data. And the data sources that I use and appreciate the most are National Medicaid Claims Data, which is a huge resource. But we should be strengthening our evidence generation with innovative linkages to new sources of data that are, in my opinion, relatively underexplored.

And so, these might include web-based, consumer facing directories, patient surveys, qualitative data, and commercial and proprietary databases which will give us more comprehensive understanding of the patient experience.

So, I'll end here. Thank you so much to NIMH for the opportunity to chat today and to all of my collaborators and mentors and co-authors and other funders. And thank you so much.

SERENA CHU: So, thank you, Dr. Zhu. Now I'd like to welcome back to this stage all our speakers, so we can have a moderated discussion and questions from our live and virtual audience. So, if you have a question, please make your way to the microphone.

JON COOPER: Good afternoon. My name is Jon Cooper, and I am director of health and wellness for a public school district in Delaware. I have a question about whether any of the research that's being done has implications for future differential diagnosis of disorders on the slide that I really liked with the low essentiality. I forget what the other term was. But that really struck me.

And I guess I just wonder if at this point there's a feeling like, "Well, you don't feel good, so we're going to help you." Versus, "You have this, you have this, you have this."

SILVIA LOPEZ-GUZMAN: I'm not sure that we're -- I think we've moved away from the idea that, you know, we're going to fully commit to the taxonomy that exists. I also don't think that we're on the other end. Right?

And I think it doesn't make sense to be fully on the other end until we fully figure out a better mapping of neural and, you know, biological mechanisms of mental health disorders and the way in which those interact with the environment, with culture, with social factors, with development, all the things we've said. So, I guess to answer your question, I wouldn't say no. We're not there yet, and I don't think we should be.

JON COOPER: Thank you.

SERENA CHU: Oh, okay. Other questions?

FEMALE SPEAKER: Hello, everyone. You know, really wonderful panel of presentations. And while I was watching -- I didn't get to see everyone's but for most -- the question in my mind was policy impact and sort of long term. And it was almost like, Dr. Zhu, you read my mind. Because you had really cleanly pointed out specific areas where you can build into that policy work.

And so, I would like to hear just from everyone, your considerations -- and even, I think, Dr. Hagaman especially in your presentation, in the very beginning, you pointed out how it was just in 2022, the decriminalization of suicide in Bangladesh. And this is a popular issue in other countries.

So, how do you see -- how do folks see their work impacting policy eventually? And/or how are you trying to incorporate that into your research right now? Are you conducting stakeholder involvement or consultations or conducting trainings with people in government or related policy work to create that change later on? Or do you not see connected at all?

ASHLEY HAGAMAN: Is this on? Okay [laughs]. So, I mean, policy in every context is so different. And so, in a place like Nepal, policy is actually determined at, like, the local city level. Like the equivalent to what would be our counties.

And so, so much of the advocacy actually that's built into our research programs are having to create municipal advocacy structures. And the youth program I was talking to you about, someone from our youth advisory board sits on the municipal government as, like, the youth advocate.

And so, we've built that in into a lot of the ways that we've been working with our stakeholders and thinking about how we make sure that it's woven into some of the systems that we work.

Other policies already exist, and a lot of our work -- Nepal is, like, so wonderful at creating policies for lots of things. And so, a lot of our work is to figure out how do we get that implemented?

So, how do we get every primary care physician well trained in suicide detection and response? And how do we do that with care and quality? So, I think we also think about, like, how are we helping the policies that already exist get deployed in meaningful ways?

SERENA CHU: Did -- Dr. Zhu?

JANE ZHU: Sure. So, I think that policy is where research should end and begin, but that's a personal bias. Because there's a lot of great research that's out there, and it doesn't get implemented in practice.

And if you talk to policymakers, a lot of them are operating blindly. There are policies that are implemented that are not based on evidence. There are policies that are implemented that are never evaluated for their efficacy, and there's no outcomes or accountability.

And so, from my perspective, mental health care is so deeply ingrained in the systems in which we practice that it is really hard to produce research without that going to the end user. Which, for me, my research is for the policymakers. It is to inform their policies so that they can do a better job at creating programs and policies that improve, you know, healthcare.

So, what I do in my research -- and this is a lot of thanks to NIMH and their, you know, sort of prioritization of advisory boards, for example, and dissemination and impact -- we have policy -- policymakers that sit on our advisory boards and are part of, you know, the research process from research generation to dissemination of results.

I sit in on meetings with policymakers all the time at briefings, discussing our results, and trying to have them understand sort of the evidence that we generate. And so, it is a very important part of my work. And I hope it's -- it becomes a more important part of most researchers' work.

KIMBERLY BOLLER: Hi. Thanks so much. I'm Kim Boller from the American Psychological Association. I so appreciated hearing about the use of many different methods, including the hybrid method and also secondary data analysis of the Medicaid data, for example.

As you're thinking about the workforce of the future and truly addressing population health issues, what are you seeing as the need for pathways and training for folks to do the kind of work that you've been doing? Thanks.

ASHLEY HAGAMAN: I think a lot of it is just inviting all the different disciplines into projects. And so, we have anthropologists and psychologists and economists that are all working on the same team together.

And so, I think a lot of it is when you bring those disciplines together, there's so much, like, new shared knowledge and, like, systems and how people think and how people work. And a lot of the ways that we try and do that is by bringing in our community members, and they have an equal seat.

And I think in terms of, like, building the next generation, a lot of what we do is just create structures within our research team for pipelines and pathways for people to grow in their careers.

So, a lot of times in low-income settings where I work, there's sort of this, like, perpetual research assistant. And we want to make sure that those bright shining stars get into leadership roles at some point. And so, we've started to create a lot of pathways for that in the teams that I work.

SERENA CHU: Yeah. Brenda, we'll take you as the last question, and then we'll go to break.

BRENDA CURTIS: From NIMH. And I have a question from a participant who's online, and it's for Dr. Rodman. What do you feel is lacking in the country's education system to support the elimination of mental health stigma? And they go on and say, starting by partnering in communication with adolescents?

ALEXANDRA RODMAN: This is a great question. In some ways, I think we've come a long way with stigma in terms of there being a lot more open conversation surrounding mental health needs.

I think a new phenomenon that, at least in my work, we're discussing with our -- actually our focus groups with our adolescents, is the fuzzy definitions now of mental health disorders and the boundaries between --

Everything exists on a continuum, right? So, it's like while we're trying to operate within this taxonomy of, like, discreet boundaries, it's not how the real-world works.

And so, I think trying to de-stigmatize mental health while still not over pathologizing has been a difficult balance for teenagers today, especially with, like, TikTok and these, like, influencers who are kind of co-opting the conversation.

And experts, I think, have difficulty kind of penetrating that conversation with more, I think, informed perspectives that may not -- that don't cause undue harm. So, I think it's a really important area of conversation.

And something I want to kind of add to that, that has been circulating my mind as some of these other comments have come up, is something I think we should think about is the system as it is -- you know, psychiatry operating within the medical model -- the system itself is quite porous for a number of reasons.

And as a nation, historically, we very much value patient privacy. But what I've seen -- this is not my research hat, this is just kind of like a person [laughs] -- a former clinician and now walking through the world, is -- what I've noticed is that, you know, at 18 years old, adolescents don't suddenly become adults, right?

You know, there's still -- it's still very much within a developmental phase. And yet our approach to care is very individualized. And I don't mean from a precision perspective. Because I very much support the precision movement, and, you know, we're hoping to do a little bit of that ourselves. But what I mean is that the adolescent-existing context, right?

And so, I think that mental health care would also do well to really think about that tension between treating the patient as a -- as an individual and respecting privacy and ethical boundaries but starting to expand care to incorporate families and other support networks that would help kind of bridge some of these gaps that just we hope -- we wish did not exist, but simply do exist.

So, we're not picking up the pieces after, you know, an episode, but preventing it in the first place. Because we all know the costs that come along with having to then recover from an episode, for example. Yeah.

SERENA CHU: Thank you. We give a hand to our guests.

[applause]

Thank you for very wonderful presentations. I just want to remind everyone that we'll break until 2:15 p.m. And just a reminder that there's no food or beverage permitted inside the theater. So, please consume all those items in the lobby. And can the NIMH Planning Committee and Event Workgroup 3, please assemble on the stage, so we can take pictures? Thank you.

(Whereupon the Subcommittee members took a brief break starting 1:55 p.m. and reconvening at 2:15 p.m.)