A Therapist in Your Pocket?: How Do Mental Health Apps Work?

Jilly Carter: Hello and a very warm welcome to this, our second episode entitled “A Therapist in Your Pocket: How Do Mental Health Apps Work?” Today, we're going to focus on the science behind digital therapeutics in the mental health space, including the neuroscience driving them and how these principles are incorporated into a digital therapy. I'm Jilly Carter. I'm an ex-reporter and presenter on television and radio, and I've worked for companies like the BBC, ITN, and WRAL-TV in North Carolina, and I'm your host for today's podcast. Now, joining me, I'm delighted to welcome two highly respected and knowledgeable guests in this space. First, Dr. Faith Gunning, PhD. Faith is an associate professor of psychology in the department at Weill Cornell Medicine. She's the director of their Institute of Geriatric Psychiatry. Faith's clinical expertise is the neuropsychology of mental health conditions. Her body of work focuses on using clinical, cognitive, and brain-based tools better to understand the mechanisms of depression and related conditions, with a focus on aging. Now for the past 10 years she's tested and developed digital and therapeutic approaches, and her overarching aim is to help make mental health interventions more accessible. Also with us today is Nicholas Jacobson, PhD. Dr. Jacobson is an Associate Professor of Biomedical Data Science, Psychiatry, and Computer Science at the Center for Technology and Behavioral Health in the Geisel School of Medicine at Dartmouth College. His work focuses on the use of technology to enhance the assessment and treatment of mental health conditions, such as anxiety and depression. Dr. Jacobson’s smartphone apps have been downloaded and installed by more than 50,000 people in over 100 countries. Very impressive. Now, we've so much of interest to talk about today. So let's dive straight in and have a quick chat about the science behind mental health apps. Now, it's my understanding that a good app is the product of a great deal of collaborative work, often with people of many, many different backgrounds. So Nick, let's start with you. Tell us a bit more.

Nicholas Jacobson: I think that's a great opening, actually a great summary of really what makes for a very effective— a likely scenario for a digital therapeutic to likely be effective. There's many companies in part in this space that go straight to market that really have folks that are not really part of the team that could likely enhance the efficacy and the safety of the digital therapeutic. So most of the applications out on the market directly available to consumers haven't been tested in any way through some type of study. And part of that is, I think, the accessibility of making an app is actually a fairly low bar and putting it out on the App Store or the Play Store, for example, is not a particularly hard feat. I think in terms of what will make the ingredients for what will make an app likely to be the most safe and effective it can be is really partnering with folks across different disciplines. So among those disciplines, folks that have experience in treating mental health are pretty valuable members of the team that are sometimes neglected in some of these early commercial strategies. So psychologists, psychiatrists, social workers, counselors, other folks, I think should be part of the team. I think at the same time, if we were to focus and wait solely on those folks, we'd be missing a big picture of the pie of what would make a really engaging application or what could be both secure and user-friendly. So I think a lot of UX and UI design, so front-facing development, full-stack development, and engineers, computer scientists, I think being part of the team is really important to try to make sure that these products are as engaging as possible. I think the other thing that is a collaborative effort that is sometimes neglected but is really important in actually trying to make sure that it matters for the end user is designing it with those folks involved as part of the process directly. And sometimes when we think about these processes, we don't think about that end of the end user design. So really constructing these applications with direct input along the way on what the goals are and how things should function with the population that's intended. And so I think all of these kinds of collaborative efforts will really enhance both the efficacy, the safety, the security, and ultimately the engagement within these platforms. Jilly Carter (6:04): Absolutely a holistic approach. It does seem, I think you've kind of insinuated that there is a glaring lack of regulation out there and there's a lot of non-experts creating these apps. Faith, it's my understanding that the FDA doesn't have to approve psychotherapy apps. So how on earth is the consumer going to be able to discern what might work for them and what, potentially more dangerously, might not.

Faith Gunning: So you are correct that that's not a requirement at this time. And I think a lot of users— we really have to, in this space, we really have to do our best to educate users about what supports the app. So from my perspective, you know, it should be very clear whether there have been studies done to support the effectiveness of the app or not when it's put out there. And that, I think, should be on the burden of the developers and the people putting the apps and making them publicly available. And also not just educating potential users but also clinicians who may recommend these to their patients, their clients. I think a lot can be done as far as educating through providers as well. I agree with everything Dr. Jacobson said about the collaborative effort. And I think another aspect of it for these apps that is quite helpful is, one thing we know that's a barrier with digital therapeutics is often people stop using them right away or they don't use them very long. And so I think having people on the team who are really experts in engagement, so for example, relying on expertise from behavioral economics and/or expertise from gaming designers. So to— the folks who are really focused on keeping people engaged.

Jilly Carter: Faith, can you tell me– I mean, the young seem to take to these apps like a duck to water. But what about older people who might be confused about this technology? Should we also be thinking about that in terms of how we design to make it as simple as possible but, in the words of Einstein, not simpler.

Faith Gunning: Absolutely. And I think a common misconception from the public is that many older adults won't be able to use these kind of apps or won't benefit from them, which we know isn't true. But adapting the apps so they're really optimized for older adults and their specific needs and aren't targeted toward young adults, because a lot of the apps out there really are targeted toward young adults. So the content may be slightly different or modified so it resonates with older adults. And also from the user experience, just presenting information in such a way that it is— it may be, you know, older adults may prefer to use ana tablet versus a phone so the print's bigger. Some of the feedback that I've heard from others and we know from folks in the community is also just adapting if things need to be not just audio but also have visual– like words printed on the screen in addition to the audio. So to get around some of the challenges that do come with aging and make the apps more user-friendly and targeted. So not just picking something up and saying, okay— that we use in young adults and saying, okay, try this out.

Jilly Carter: Which chimes completely with your whole passion for accessibility. So let's turn our attention now to the theory behind how these therapies change the brain, hopefully in a positive manner. We're very comfortable with the term mechanism of action when we talk about traditional medicine, but can you both elaborate a little bit on how these apps are different when it comes to mechanism of action. What might be happening in the brain, Nick, and which pathways are they targeting?

Nicholas Jacobson: I think that's a great question. I will defer to Faith on the neuromechanisms of these, but I will at the same time highlight that this science— although there are some new ways in which we are targeting things, so there are certainly some mechanisms that are explored in digital therapeutics, which are really large and wide class of different activities that can be grouped there. Many of the strategies that are targeted within digital therapeutics are really just different delivery vehicles for content that has been existing and developing and studied intensively over the last half century or so pretty rigorously in terms of within the context of psychotherapy. So I think the new delivery vehicles for how these things can be deployed, the level of access over time, I think for me, I'm particularly interested in potential passive behavioral markers of actual changes that happen during the course of our treatment as a potential mechanism— a behavioral mechanism of action. So for example, one of the things that we are often targeting in major depression is this reduced level of activity and withdrawal from daily life. And we can study a lot of that through our smartphones, for example. So a lot of what we do is try to measure folks withdrawing in different ways and not participating in other aspects of their life to the same degree when they have increased levels of both depressive symptoms or meet diagnostic criteria for the disorder and thinking about that as a potential marker for something that we are directly actually targeting often within these digital therapeutics and see if we can actually measure behavioral changes. So are folks calling, texting their friends more often? Are they literally getting out and about from the house? These kinds of things can be something that we measure passively and could be an indicator of actually what would be a mechanism of action in terms of the behavioral realm. I think there are corresponding impacts, as you've alluded to, on neurofunctioning, and I think Faith is an expert in that, and I will defer to her on ways to really elaborate on that.

Jilly Carter: Go for it, Faith.

Faith Gunning: Thank you. So as Nick said, so the science here for psychotherapy, the behavioral mechanisms of action we do think and have evidence to believe are quite similar between traditional psychotherapy and digital therapeutics for mental health. And the evidence from the neuroscience, from the brain standpoint and the networks that are involved, is quite preliminary compared to other areas of neuroscience. I think two areas, to pick up on what Nick was saying about behavioral activation and increasing people's behavior, there is some preliminary evidence, we know this from traditional psychotherapy, that behavioral activation, if somebody increases their rewarding behaviors, that this changes reward circuitry in the brain, which is really important for— because that is a deficit, an area of weakness or what we think of as dysfunction deficit that we often see in individuals with depressive disorders. Another area that I think has, even though the work still is preliminary, that has— has received more attention with digital therapeutics is targeting what we call executive control, which is, it's a cognitive function, sort of a broad cognitive term talking about, you know, planning, the ability to plan, organize, multitask, these functions that are really important for daily life and that are often disrupted in individuals with mental health conditions. And so there is some evidence that digital therapeutics that are targeting this executive control network can— do change the network and improve its connectivity with use of the therapeutic. But this area is pretty new and I think will evolve pretty rapidly. And some of it we're still extrapolating from the traditional psychotherapy literature.

Jilly Carter: Can you, Faith, expand a bit on this thing called digital biomarkers for those who are not quite sure what all this means? Can the effect of an app be seen or measured on an MRI scan, for example? Can it be measured in the real world?

Faith Gunning: So there is evidence that the effect of a digital therapeutic can be measured using MRI scans. And these MRI scans that have been done within this context are really focused on synchrony of activity among brain regions and the organization of brain regions and their function. So there is some evidence that the brain does change with the use of some of these digital therapeutics. The digital therapeutics, to my knowledge, that have been tested thus far are more narrow within the context of those kind of digital biomarkers. And I am going to defer to Nick about other potential biomarkers that I think he has focused more on.

Jilly Carter: Yeah, Nick, tell us more.

Nicholas Jacobson: Sure. So I guess, one, that term can be used for many different things. So I think it's a little hard to define exactly. The term is often commingled with digital phenotyping to some degree, which is the individual quantification of moment-by-moment changes in our behavior and physiology based on our everyday devices. So sometimes these things are used in that way. They are used differently by different people, but that's a term that I would generally put them in that umbrella myself. In terms of, like, an example of that, the types of data that we could use to try to form what is a potential marker is some of the data that I've already talked a little bit about. So the things like the types of behaviors that we engage with in using our applications, for example. So using our phones could give us a sense of, really, our day-to-day lives in ways that could be a marker of some state related to mental health. So I do think that I like, actually, the term— the use of the term marker in large part because we don't know if they're— a lot of what we're trying to do is see if these different types of signals, so GPS mobility, so how much folks are getting out and about, what types of locations they're visiting, how they're spending their time— there are different types of contact both in person, so turning on the microphone and trying to detect whether they're in conversation with other people. Things like their level of activity, the physical activity duration, and also markers of sleep, these are all just examples of things that we can infer that are both related to some degree to both our physiology and our behavior and that could be— we usually are using in a predictive way. So we were trying to use that type of data to try to predict some type of mental health outcome. And we don't know if they are necessarily, when we're trying to predict those things, causal relationships, but we do have really good evidence that these types of things can actually be a marker for both severity and then course for different types of mental illness. And so I think that that term is decently appropriate given the type of evidence we're evaluating. And we are interested in these as potentially putative causal mechanisms, but that requires a lot more to get into cause and effect. So I like the term marker in a lot of ways.

Jilly Carter: Do digital therapeutics that incorporate elements of neurobehavioral techniques have the same mechanism of action as face-to-face treatment? I'm not quite sure about that.

Faith Gunning: Faith Gunning So I'll start. So I think that some of the mechanism of action from what we know is certainly going to overlap. And Nick already talked about some of the behavioral mechanisms of psychotherapy. I think the one area, depending on the digital therapeutic, but for the digital therapeutics that don't involve any— that are standalone and don't involve any individual therapist or coach or anything, I think that's where some of the mechanisms of action would likely differ because in traditional psychotherapy, there is a powerful effect of— often of the relationship with and the alliance with the psychotherapist. So that part is not— it's not part of many digital therapeutics.

Jilly Carter: Anything to add on that, Nick?

Nicholas Jacobson: Yes, absolutely. I think one of the things that is really strange in getting into this world is, in part, the attachment that folks do actually develop with these tools, even when they have no actor. So there's a term within the literature that is just called the digital therapeutic alliance, which is really taking the way— the main ways that we conceptually would study the relationship between a patient and a provider and the ability to get on the same page and develop and trust that relationship to work on a common goal. There is some of that that actually happens with these inanimate software products that to some degree actually seems to, like early evidence points, even in the digital world, that is a good, important facilitator of successful outcomes, which is— it's strange to me that that kind of, there is some mirroring in the literature, but it's quite different in terms of applications. These are nothing like often interacting with a human. I think the main thing that is the wild card that makes them a little bit more— potentially more similar is the use of generative AI, where things can be far more human-like in their application. And that's a very new field. So the science on some of that is really not there yet. We've done some early work in that area for a while now, but there's still a lot yet to learn as to how well these types of things actually translate in that kind of area.

Jilly Carter: One word you mentioned, which I'd like to pick up on, is trust. And if you're monitoring a patient remotely, the question of trust must come up a lot. And clinicians and such like will be privy to very sensitive information. And patients might understandably feel a little uncomfortable about that. So how do you address this issue? Because trust is essential.

Nicholas Jacobson: Yeah, I think that's a really fantastic question. I think that's at the bedrock of all psychotherapy, as you alluded to. How do you do that well? I think you'd be open and transparent about what you are doing and why and what— in our case, for example, the data that we'd be collecting on folks from a passive data perception. We need to make sure that we are very open about what we are collecting, why, and how that can really be beneficial to them, so like why we would collect this, why that's— and when we're doing work related to the use of, like, actually trying to use that as part of the digital therapeutic, how this may be helpful for us to collect that for their own self, their own self-awareness. So that's a piece of developing that is getting some value. It's not just a give. It's a take from the— from that perspective to some degree. But at the same time, I think one of the things that’s— I'm— we have ways of trying to make sure that folks are aware of that within the teams and within academic work. So we, like, literally quiz people to make sure that they know what data are being collected and things like that following when they would agree to a consent form and actually sign it. We actually ask them comprehension questions to make sure that, especially the sensitive areas, that they really do know that we are collecting these things. I think as we move into, like, translating this from academia into industry, this— the motives of this, I think, can drive some outcomes in some areas that would be different in terms of— terms and conditions are pretty easy to tuck away in a little check mark that you don't necessarily have to read, for most folks, or most folks won't actually read them rather. And so I think this level of transparency, we kind of hold ourselves to this, and IRBs hold us to this. But as we translate into industry, the incentive structure is generally to try to— if we were going to try to translate this, not all industries would be equally transparent. I think there's actually a number of analogs that have been happening in— outside of the mental health area, and well, actually, there are some examples in mental health. But outside of mental health, it's fairly prevalent for a lot of these types of sensor data to be collected without the user really knowing that these are often going to data warehouses and being sold as an aggregated en masse to— for example, GPS location data is done fairly commonly, so. And I think that's a real challenge in terms of the lack of regulation surrounding a lot of this activities and folks to be on the same page. I'm far less concerned about these areas in academia where there's incredible over— levels of oversight both within teams from themselves and but also by institutional review boards that are— will check over and review every protocol we have. But in industry, you know, I think that these kinds of things do come up, and I don't have the answer for how an individual should navigate this because it's, in part, is like trying to read every terms and conditions that you might ever need to read, I think, frankly, is just a huge time commitment and impactful.

Jilly Carter: Now, you've both been in this field for a long time, and you both have deservedly high reputations on the research front. Can you share with us how your research has gone on to be integrated into digital therapeutics and what the successful outcomes have been? Faith, do you want to start?

Faith Gunning: Sure. So I'll start. So some of— my main area of research that coincides with also my interest in digital therapeutics or led me to this interest is, much of my research is focused on understanding who doesn't respond, like, to our traditional interventions, whether that's pharmacotherapy, often medication, or psychotherapy. So what we've done is we've studied individuals who don't, for example, who don't respond well or get— or improve as much as we would like to see or they would like to see with the traditional antidepressant medication. And then we have used that to— our understanding of what makes somebody not respond and what can we do within this space to really target and optimize our interventions for individuals who are left symptomatic with the most available treatments for them. And so we've done that. I mentioned the main area that we focused on is individuals with this— who tend to have this— who have this difficulty with what we call executive control often don't respond to our medications that are typically used. So we've targeted that group and optimized our interventions in order to treat both the mood symptoms and the cognitive symptoms that are often disrupt daily functioning. And then the other area that we've become focused on more recently is, we know from a body of work now, and this seems pretty obvious, that people— the level of people's social engagement is tied to mental health. And if we know that individuals who feel socially disconnected don't respond to traditional psychotherapies as well either. So we've really taken this finding that we've seen over and over again and others have seen as well. And now some of the therapeutic approaches really are targeting and not just increasing people's behavior, but getting people to increase their rewarding social interactions, which seems like a bit of an oxymoron that we're doing that using digital therapeutics. But that is how we're helping to deliver that to people in the community. So those are a couple of examples of how we've integrated. And this is work I've done in collaboration with a large group.

Jilly Carter: Yeah, absolutely. And Nick, what are you finding? How your research is really making a difference in terms of these digital therapeutics?

Nicholas Jacobson: Yeah, I think in terms of most of what we design, we try to design what we are doing to be exceptionally cost-effective for every new user, in large part because why I got into this field is to directly have a potential to make a difference with the work that we do. Release it out on the open market. And we have had some success in getting the things that we are designing out into the world with a lot of folks that have actually gone on and used it to try to change and manage their symptoms. So I think that's one way that we think about this is direct deployment of these things. I think another way that we are doing this is partnering with folks that are in industry. I think some of our most successfuls have actually been in partnering with non-for-profit organizations that are really trying to disseminate this. So we've done work related to trying to target, for example, body image on social media. And that's actually directly deployed on social media. It continues to be live and is maintained by the nonprofit organization targeting folks that are searching for specific terms that might put them at risk, for example, and being offered some of this. So I think there's a lot of different ways to translate it. I think often partnering with an industry is another major way that a lot of folks in the field have impact, or creating companies themselves. And there's, I think, a great need for partnerships between academia and industry to try to make sure that what is deployed is more evidence-based, and folks that are really in the weeds studying these things actually make it out. So I think there's a large role these academic industry partnerships to try to make sure that what makes it out is as efficacious as it can be from the start, but then effective as it's delivered across all of these different populations and maintained by a company.

Jilly Carter: Sure. Now, given that it could be described as tiger country out there with precious little regulation, as we said at the start, do you both believe that there should be specific types of studies carried out to prove the effectiveness of each digital therapy? Because at the end of the day, it's about results in the real world and helping patients, as you've said, live their best lives.

Faith Gunning: So I'll start. I do think what we know is that of the studies that are done, a lot of them are done in more controlled settings, which is, in my opinion, the place to start often is starting in a very controlled, like randomized controlled trial that you have a lot— the team has a lot of oversight over. And then moving from there into more community settings, whether that's through social service agencies, school systems, doctors’ offices, but then moving to the community. And that's where, I think— and that's where I think we need more work, high-quality work done within these community settings and partnering with community agencies or providers who really are, like— we're, like, partnering with social service agencies and also partnering with community providers. Nick, I know you have experience in this area, so.

Nicholas Jacobson: Nicholas Jacobson: I think that one thing that I kind of think about a lot is what, how, when we design these types of studies, what the control should be. And so when we're testing whether something is effective, it's compared to what. And to some degree, I think it depends a lot on what the intended deployment vehicle for where this would actually live as to what the best control might be. So if you're going to try to replace a clinician, which I wouldn't, you would want to compare that to probably actual clinical service, like routine care as is delivered. But if you were to try to go into a market that is where folks probably aren't receiving treatment at all, I think that that's most— like, to some degree there, you could argue that two controls are probably most appropriate. One is nothing at all, in which case they're likely not receiving it. The other is another— another digital therapeutic or something that they could be testing themselves or made available also direct to consumer. So another application that they could download through the market, depending on how you're finding these folks. And I think to some degree, the— most of the time I think about, because most people don't receive any type of specialty care for their mental health, in a lot of ways, I think to some degree, the best control, at least as a starting point, is actually against a weight like folks that are just waiting to have access to that digital therapeutic, not receiving another active intervention. This is a controversial thing in the field, and folks have very strong opinions about what the control should be. But I think there is some good reason to think about comparing it at least to what most folks receive, which is nothing.

Jilly Carter: Right. So as Nancy Klein said, for a difference to be a difference, it has to make a difference. So what success stories could you share with us in terms of patients leading fuller, more fulfilled lives than they previously endured?

Faith Gunning: So I will say that some of the people we have seen through and used digital therapeutics with, we have had some folks who were able to go back to work and who were not working because of their careers were interrupted. So I think that that is a very obvious impact. Other areas, we've seen people who have become more engaged with the significant others in their life, as we know often that social isolation comes with many mental health conditions. Going to class, those kinds of, like, I think those real world— not just, like, seeing a change, seeing a change on measures is good, but we also like to see and have seen people's daily functioning improve, becoming more engaged in their life again. And again, it seems like to many, it seems like an oxymoron because they think, oh, how could you, people spend too much time on their tablet or their phone, you know. But the goal, it's— we don't— we also like to, you know, we're not advocating for people to spend hours and hours a day.

Jilly Carter: Nick, anything you'd like to share in terms of success stories?

Nicholas Jacobson: Yeah, I think one of the ones that in terms of functional outcomes that I think are just so fun and also kind of astounding to see are the changes in real-world behavior. I think one of the things that I love about passive sensing is we can observe them, so we actually see directly based on behaviors folks, like, increasing their social communication with other people, and that's really, like, as we're providing these interventions compared to a control group, that's really exciting work to see that, like, these change naturalistically in the course of their daily lives based on them engaging with these things. I think there are other individual meaningful outcomes. So folks that are really going through and struggling with this social withdrawal that will start to engage again in the world. And I think one of the things that is just really exciting about this is there are a lot of those kinds of individual stories, but the delivery mechanism of these types of things is so much more scalable that, in a lot of ways, when I think about these things, these individual narratives are part of this huge collective whole that I think is really exciting. So I think when we think about this work, it's not just one of those that we might have with— if we were a clinician treating someone over the course of a couple of years that this happens, we could have a large number of folks that were going through this at the same time engaging with the tool, which is part of the excitement around the potential for these things to impact a large number of people simultaneously and are not bottlenecked by clinician time. Which is part of what I think is really exciting about, like— it's almost to some degree, these individual narratives are really exciting. But when you start to think about many individual narratives all coming together, it's even more exciting in my mind.

Jilly Carter: We've covered a lot of topics here today. Thank you both so much for your insights and your knowledge and your experience that you've shared with us over this past short time together. Thank you both so much for contributing. And thanks to everybody who's been listening. We hope you found it uplifting. Thank you very much and goodbye. SC-CRP-16902 | SC-US-78120 | 12/24