Building digital mental health approaches to prevention and early intervention

Transcript

• 00:00Good afternoon, everyone,
• 00:03and welcome to Grand Rounds.
• 00:04Please enjoy the coffee and
• 00:06take your seats when you can.
• 00:08And it's a pleasure to welcome you
• 00:10back here to the Cohen for grand rounds.
• 00:12And I'd like to just start with echoing
• 00:14the sentiment expressed by Linda and
• 00:16Tara and their message last week.
• 00:19And I want to apologize on behalf of
• 00:21the entire Grand Rounds Committee
• 00:22for scheduling grand rounds when
• 00:24some members of our community
• 00:25are observing Rosh Hashanah.
• 00:27And sure, I would like to wish everyone
• 00:29who is observing Rosh Hashanah.
• 00:31Uh, Shana tova, if you're watching
• 00:33this back on the recording,
• 00:35I hope you'll take us up on our
• 00:36offer to schedule a virtual meeting
• 00:38with Doctor Allen and to connect
• 00:40with him about his research.
• 00:41And I'd like to thank Doctor Allen for
• 00:44making himself available to meet with
• 00:46interested faculty later on this week.
• 00:48Now, just a reminder.
• 00:49Next week we will be back here in
• 00:52the Cohen Auditorium for our first
• 00:54compassionate care rounds of the semester,
• 00:57led by Laurie Cordona, Dr Laurie Cordona.
• 00:59So please do join us for that.
• 01:01And then.
• 01:01So all of our sessions.
• 01:03Our ground rounds sessions for the month of
• 01:05October will be in person here in the Cohen.
• 01:08So join us for a coffee as we come
• 01:10together and as we learn together.
• 01:12Now today it's a great pleasure
• 01:14to introduce our speaker,
• 01:15Doctor Nicholas Allen.
• 01:16And so Nick is the founding director
• 01:19of the Center of Digital Mental Health
• 01:21at the University of Oregon where
• 01:22he's the Ann Swindells professor in
• 01:24clinical psychology and Nick trained in
• 01:26clinical psychology in the University
• 01:28of Melbourne with postdoctoral
• 01:29training here in the United States.
• 01:32With Peter Levinson and before
• 01:33returning to Melbourne to take a
• 01:35faculty position where he worked for
• 01:37over 10 years in the National Center
• 01:39of Excellence in Youth Mental Health
• 01:41before returning to the University of Oregon.
• 01:44And as many of you will have
• 01:46read in Nick's research,
• 01:48he applies innovative digital technologies
• 01:50both wearable and mobile to try
• 01:53and predict or prevent suicide and
• 01:55to better understand risk states in
• 01:57the context of adolescent depression.
• 01:59He's disseminated his research
• 02:00widely in journals.
• 02:02Including in Nature in PNAS.
• 02:05And in 2019, he founded Kasana Health,
• 02:08which is a company whose mission
• 02:10is to accelerate the translation of
• 02:12digital technologies to improve the
• 02:14care of individuals mental health.
• 02:16So, with no further ado,
• 02:18I'd like you to join me in welcoming Dr.
• 02:20Nick Allen.
• 02:29Well, thank you, Kieran,
• 02:30and thank you everyone for making
• 02:32time to come and listen today.
• 02:33I really do appreciate it and let me
• 02:37just get my share screen working.
• 02:43There we go. Good. And it's
• 02:47great to be here at Yale.
• 02:50I, as you can tell I did
• 02:52not grow up in this country.
• 02:57So I have a competition with my wife
• 03:00about which states we've visited,
• 03:02and the definition of visiting a state
• 03:04for the purpose of this competition
• 03:06is that you have to have slept there.
• 03:09And I just ticked off Connecticut.
• 03:12So thank you for that.
• 03:16Anyway, I want to talk to you about
• 03:18our work on adolescent depression
• 03:21bringing in digital methods.
• 03:23Hello? Why is it not responding?
• 03:28There we go. All right.
• 03:30We'll start with the disclosure,
• 03:31which you've already heard,
• 03:32which is that I have founded a
• 03:34company called Kasana Health,
• 03:35a digital mental health company,
• 03:37and I have an equity
• 03:39interest in that company.
• 03:44Now this is not necessarily a
• 03:46smart thing to do, but there's this
• 03:48website and what you can do on this
• 03:50website is you can go and pick a
• 03:52particular date and it will give you.
• 03:54Data on the burden of disease for
• 03:57different categories of disease
• 03:58on different dates and so I.
• 04:03In a crazy experiment,
• 04:04I went and looked at this,
• 04:05and I and I put in the age I was when I
• 04:07graduated with my PhD in clinical psychology.
• 04:10And then I put in today's date.
• 04:13And what this showed is that although
• 04:16we've made really great progress in many
• 04:18areas in medicine with things like this,
• 04:22particularly infectious diseases.
• 04:25Obstetric issues and so forth.
• 04:29Mental health problems have not shifted
• 04:31at all in terms of the burden of disease.
• 04:34Nothing, not a jot,
• 04:36and to make matters worse.
• 04:39I have spent my entire career
• 04:41focused on mental health and youth,
• 04:42so adolescents and young adults.
• 04:45And as you can see,
• 04:47this is a stage of life where the
• 04:50burden of disease of mental disorders
• 04:53is at its peak across the lifespan.
• 04:56And things are actually getting worse.
• 04:59We think for these, this group of people,
• 05:01so you can see here some data on depression.
• 05:07Symptoms across across time from
• 05:09two different studies showing that,
• 05:11particularly for.
• 05:12Young girls,
• 05:13there's an increase in these problems,
• 05:16particularly over the last decade
• 05:18and 1/2 to two decades.
• 05:20And moreover,
• 05:21suicide is increasing in this country,
• 05:24not all countries,
• 05:25and has overtaken homicide as the
• 05:28most common as as a as a more common,
• 05:31sorry form of death amongst young people.
• 05:34So.
• 05:37So what have I been doing all this time?
• 05:40I've been working on the problem
• 05:41and we're not making progress.
• 05:43So that's that's a that's a concern.
• 05:48Now part of the problem is
• 05:50we know it's a hard problem.
• 05:52We know it's a wicked problem, so.
• 05:55And one of the difficulties is it's hard
• 05:58to know how to scale up our efforts in
• 06:01a way that will really have impact.
• 06:04So this diagram here comes from a
• 06:06report that I was involved in for the
• 06:08World Innovation Summit for Health,
• 06:10so on digital technologies and mental health.
• 06:12And what this this graphic points out is
• 06:17that if you calculate it a certain way,
• 06:19you could argue that the current mental
• 06:21health system is not effective for 90% of
• 06:23the people with a mental health problem.
• 06:25The way you get there is you say
• 06:28OK for every. 10 people in need.
• 06:32Only four people actually access treatment.
• 06:35And for those,
• 06:36only about 1.5 of those people
• 06:38will access what we would call
• 06:40minimally acceptable treatment,
• 06:41so treatment of adequate quality.
• 06:44And evidence base.
• 06:45And then even though evidence based
• 06:48treatments will get variability and outcome,
• 06:51So what this highlights is that we have
• 06:54a number of kind of grand challenges
• 06:56in the field that we need to solve if
• 06:58we're really going to bend the curve,
• 07:00this curve that seems to be
• 07:02bending in the wrong direction. So.
• 07:05We need to think about prevention.
• 07:08So how do we prevent these people
• 07:10from getting into that first group
• 07:12in the 1st place?
• 07:13We need to think about access
• 07:15to mental health services.
• 07:17We need to think about the quality of the
• 07:19mental health services that people receive.
• 07:22And we also need to improve the
• 07:24effectiveness of our services.
• 07:25So this is a nomenclature
• 07:27that I'll come back to,
• 07:28but I think is really critical in terms
• 07:30of thinking about how we solve the
• 07:32problem and start to make progress.
• 07:37So. My own work has been largely
• 07:42informed by throughout most of
• 07:44my career by an an interest in
• 07:47prevention and early intervention.
• 07:49And of course, we know that the mental
• 07:52health system has traditionally been designed
• 07:54to deal with chronic and end state illness,
• 07:58to respond to crisis rather than
• 08:00to be preventative in orientation.
• 08:04And so, so this is a a paradigm shift.
• 08:06It's occurring in many contexts
• 08:08around the world.
• 08:09But this is the one that I've been
• 08:11interested in and the and I guess
• 08:13if you if you if I think about my
• 08:16program of research and clinical
• 08:17activity over most of my career,
• 08:19you know the argument that I would make
• 08:21in my grant applications and things
• 08:23like that would go something like this.
• 08:25We need to understand what these
• 08:27pre morbid indicators of risk are.
• 08:30We need to have practical and
• 08:32scalable methods for monitoring
• 08:34these and screening people.
• 08:36We need to have modifiable factors,
• 08:39so just because something is a
• 08:42predictor doesn't mean it's modifiable.
• 08:44And and then we have to have effective
• 08:46and scalable methods of intervention,
• 08:48early intervention or prevention and
• 08:50also it's been very much informed by an
• 08:53interest in developmental inflection points.
• 08:54So are there ways to target our intervention
• 08:58efforts developmentally so that we're
• 09:01getting a greater impact of the,
• 09:03so that we're meeting the,
• 09:04the plasticity,
• 09:05if you like,
• 09:06of the process with an intervention
• 09:07that's well timed?
• 09:11And I want to give you and you
• 09:12know we did we did lots of work.
• 09:13We, I've, I've run a couple
• 09:16of large longitudinal studies
• 09:18that have looked at a range of
• 09:20different risk factors including.
• 09:23You know, brain biomarkers have
• 09:25obtained through neuroimaging,
• 09:27genetics, neuroinflammatory processes,
• 09:29neuroendocrine processes,
• 09:31puberty as a developmental process.
• 09:33We've also done a lot of work on
• 09:36family interactions where we've used
• 09:38micro social observation of family
• 09:40interactions in adolescence to try and
• 09:42understand patterns of interpersonal
• 09:43relating and how they react.
• 09:44And so we've, we've we've looked at lots
• 09:47of different risk factors and there's
• 09:49many things I could say about all that.
• 09:52But, but we were searching for this
• 09:54thing that was a really strong predictor
• 09:56of outcome and that was modifiable.
• 09:58And so that's when.
• 09:59And so as a result of this we
• 10:01got quite interested in sleep.
• 10:03Because sleep is is a potent marker
• 10:06and it is modifiable so as and it's
• 10:11also developmentally sensitive.
• 10:13So particularly in early adolescence
• 10:15as many of you will know,
• 10:16there's a shift in the timing
• 10:19of the circadian process.
• 10:20Such that it's it's pushed later.
• 10:23And so there's less sleep
• 10:25drive early in the evening,
• 10:26there's more sleep drive in the morning.
• 10:28And that combines with various
• 10:32lifestyle factors such as homework.
• 10:35Sporting activity, but also the big one,
• 10:39social media, right? Being online.
• 10:41The fact is that your peer group is now
• 10:44available to you online almost 24/7.
• 10:46It's a very.
• 10:48Historically unusual.
• 10:50Context.
• 10:53And then we've got early
• 10:55start times for school.
• 10:57So it's very easy for adolescents
• 10:58to build up quite a lot of
• 11:00sleep debt during the week.
• 11:01And then they engage in catch
• 11:03up sleep on the weekend.
• 11:05So they sleep in catch up.
• 11:08But of course, then they shift
• 11:10the circadian oscillator further,
• 11:11and then when Monday comes around,
• 11:12they've got what some people
• 11:14call social jet lag. Right.
• 11:16So they feel awful because they're
• 11:17trying to reset the circadian office.
• 11:19So even though they've dealt with
• 11:20the sleep debt and the sleep drive,
• 11:22now there's circadian system
• 11:23is telling them hang on.
• 11:25This thing is getting up at 7:00 AM
• 11:27or whenever it is on Monday morning.
• 11:28That doesn't feel good.
• 11:34So this is a known issue in adolescence.
• 11:38But the other thing we know about it is
• 11:40that it's related to a lot of stuff,
• 11:41a lot of outcomes.
• 11:43It's a broad transdiagnostic risk factor
• 11:45for all sorts of difficult outcomes.
• 11:47Depression, suicide, substance use,
• 11:50cardiovascular disease later in life,
• 11:53or risk factors for cardiovascular
• 11:55disease that emerge during adolescence,
• 11:57neurocognitive functioning
• 11:58in school performance.
• 11:59I won't belabor the point,
• 12:00I'm sure you get it,
• 12:02but this is a predictor both in
• 12:04a distal and in approximal way.
• 12:05There's some very interesting
• 12:07studies that have.
• 12:08Been done on day-to-day variability
• 12:09and sleep and how it correlates to
• 12:11day-to-day variability in some of these.
• 12:13Phenomena.
• 12:16We also know from experimental
• 12:17studies such as this one from
• 12:19Allison Harvey's group at Berkeley,
• 12:21that by experimentally manipulating sleep
• 12:25you can actually see effects on mood.
• 12:29Pretty much immediately.
• 12:30And there's A and there's a
• 12:32developmental sensitivity to it.
• 12:34So if we look over here,
• 12:35let me see if we get the cursor to wake up,
• 12:38you can see that the early adolescence.
• 12:40Are experiencing the much stronger
• 12:43effect of the sleep deprivation
• 12:45than the late middle adolescence.
• 12:48So early adolescence is a
• 12:51sensitive period developmentally.
• 12:53So. We're in business facts.
• 12:56We've got an early risk factor
• 12:58that's developmentally specific.
• 13:00It's modifiable.
• 13:03Well,
• 13:03here's a bit of data that we
• 13:05still haven't published yet.
• 13:06It's in the process,
• 13:07but this is also looking at an individual
• 13:09difference called chronophage,
• 13:11so whether you're a night owl
• 13:13or an early bird.
• 13:14And what what this I want drag
• 13:16you through this complex graph,
• 13:18but what it's basically showing
• 13:20you is that the chronic phase
• 13:22prospectively this is a four
• 13:23wave longitudinal study and the
• 13:25chronic phase is prospectively
• 13:27predicting changes in depression.
• 13:29On two out of the three waves,
• 13:31depression doesn't predict
• 13:33chronic phase on any wave.
• 13:35So we also see that's a
• 13:37prospective market if you're a,
• 13:38if you're a an evening phenotype,
• 13:41then you're probably going to have a
• 13:43a a worse version of that problem I
• 13:45talked about before of the delayed
• 13:47sleep and then the sleep debt
• 13:49and then the the social jet lag.
• 13:53So.
• 13:55We did this, we developed an intervention.
• 13:58And we and we,
• 14:00it's called the Census project.
• 14:02This was done when I was in Australia.
• 14:05And we had a we ran an RCT,
• 14:08we went into high schools,
• 14:11what's called high school in Australia.
• 14:13I should point out in Australia High
• 14:15School refers to year 7 to to 12.
• 14:17So that's what what it was actually
• 14:19more what would be called middle school
• 14:20in the most of the US and we compared,
• 14:23we developed,
• 14:24we it was a group delivered intervention.
• 14:27We had one that was asleep
• 14:28intervention and the other one
• 14:30was a study skills intervention.
• 14:32So we tried to build something
• 14:34of a plausible placebo.
• 14:36For the intervention,
• 14:37we were very careful to not
• 14:40communicate to people what the.
• 14:43What the study was all about.
• 14:45And and.
• 14:47In general terms,
• 14:49it worked. We were able to see
• 14:51that it actually improved sleep.
• 14:53So this is a there's a self report of
• 14:56sleep and you can see that there are
• 14:59variables and you can see that the.
• 15:01Global sleep problems were lower in the
• 15:04intervention group sleep onset latency.
• 15:05So how long it takes you
• 15:07to fall off to sleep?
• 15:09Shorter in the intervention group
• 15:11and daytime sleepiness slower.
• 15:13We also see that with objective data,
• 15:15so this comes from wrist actigraphy.
• 15:18And we can see that once again,
• 15:19the sleep onset latency was shorter,
• 15:22there was less,
• 15:22and there was less variability in sleep.
• 15:24So kids became who in the
• 15:26intervention group became more
• 15:27regular with their sleep behaviour.
• 15:29So less variability and sleep onset latency,
• 15:32less sleep efficiency variability
• 15:34and less variability in bedtimes.
• 15:36So that's so good.
• 15:38We also find that it.
• 15:41They were less anxious.
• 15:44They experienced less pre sleep
• 15:45arousal or anxiety and they and
• 15:48they knew more about sleep.
• 15:49Full disclosure, the intervention
• 15:51did not impact on depression.
• 15:53So we didn't find an effect there.
• 15:56So. OK. We've done it.
• 16:03Take right we found a modifiable risk factor.
• 16:07We developed an intervention. We tested it.
• 16:09We published it in good journals.
• 16:14But this was a career crisis for me.
• 16:17And I'll tell you why.
• 16:19Because I could not give it away.
• 16:23How do I get it out there?
• 16:27So the problem was, as I said,
• 16:29we had found an effective approach,
• 16:32potent premorbid refactor,
• 16:33a modifier etiological factor,
• 16:35and effective measure of intervention.
• 16:36We developed it.
• 16:38It had a plausible
• 16:39developmental inflection point.
• 16:40But what about practical methods
• 16:42of screening and monitoring?
• 16:44It was so much effort to go into
• 16:46these schools and to screen the
• 16:48kids and to find them and then to
• 16:50get them to come to the groups.
• 16:51I had, you know we had millions of
• 16:54dollars of funding for study from
• 16:56the Australian NHMRC and and and it
• 16:59was tough and people worked hard
• 17:01and you know like, so that was it.
• 17:03And then secondly,
• 17:04how can we deliver this intervention
• 17:06at scale?
• 17:06It's a group delivered intervention
• 17:08making those scheduling the intervention.
• 17:10It's kind of about,
• 17:11you know the whole thing.
• 17:12So so I was sort of like I felt
• 17:15like I'd kind of done what I said I
• 17:18was going to do and and I you know
• 17:20I was still frustrated it wasn't
• 17:22having the impact and so that's
• 17:24when amongst other things I started
• 17:26to get interested in in digital.
• 17:28So and and hopefully you'll see why.
• 17:31Partly because I was focusing
• 17:33on adolescence and adolescence
• 17:34obviously used their phones and
• 17:36and other digital devices,
• 17:38but especially their phones.
• 17:41Extensively, right.
• 17:42So we've got this adolescence
• 17:43is an interesting point.
• 17:44We've got this combination of factors.
• 17:47We've got the emergence of onset of
• 17:49many forms of mental health problem,
• 17:51particularly depression.
• 17:52Eating disorders,
• 17:53substance use disorders later in adolescence,
• 17:56various forms of early psychosis.
• 18:00And other problems,
• 18:02you've got intensive use of mobile
• 18:05computing for a particular purpose.
• 18:08Social connection. Right.
• 18:11It's a device of social connection,
• 18:13and there's a developmental
• 18:14reason why people at this age
• 18:16are so interested in this device,
• 18:18because it is a is a tool
• 18:20of social connection,
• 18:21and we've got high plasticity
• 18:23and learning going on as well.
• 18:26So I've mentioned this why is
• 18:28technology so compelling from
• 18:29a developmental point of view?
• 18:31Well, it fits with a lot of the
• 18:33developmental tasks of adolescence.
• 18:34It's about connecting socially.
• 18:36You can experiment with identity.
• 18:39So you can have, you know,
• 18:40as you might know,
• 18:41kids often have different accounts
• 18:42for different aspects of their
• 18:43identity that for instance,
• 18:44and the sisters and the, you know,
• 18:46that sort of thing.
• 18:46So you actually can experiment with
• 18:48what you want to present to the
• 18:51world and to different audiences.
• 18:52You get a lot of peer based
• 18:54information and feedback.
• 18:55It's private. Especially from parents.
• 18:59Adolescents do not care much about
• 19:02what Google knows about them.
• 19:05But they care a great deal what
• 19:07their parents know about them.
• 19:08Right.
• 19:08And the interesting thing is,
• 19:10when I was young, it's a long time ago,
• 19:12and I wanted to call up my girlfriend.
• 19:15I had to talk to her parents first.
• 19:18It was absolutely mortifying.
• 19:20And so,
• 19:21and they would have to get through,
• 19:22run that gauntlet to get to talk to
• 19:24her because there was one phone in
• 19:25the house and it was connected to the wall,
• 19:27right?
• 19:28And so this idea that I could have
• 19:30directly spoken to her without the
• 19:32parents having any clue would have
• 19:34been absolutely mind blowing, right?
• 19:37So that's pretty important.
• 19:39And of course there's a there's
• 19:40a literature on board and
• 19:42pronouncing adolescence,
• 19:42which is also relevant because
• 19:46it's a divisive entertainment.
• 19:48Now The thing is, this is 1.
• 19:50This is a picture for the real old folks,
• 19:52but.
• 19:55The. The interesting thing about this device,
• 19:58though, is that it's full of sensors.
• 20:01It's collecting information constantly.
• 20:05And you don't have to do a damn thing
• 20:08to get people to contribute because
• 20:10they're contributing just with their
• 20:12naturalistic use of the device.
• 20:13And of course, this is exactly why Google
• 20:15and Facebook are profitable companies,
• 20:17because what they do is they use that
• 20:19data to target people with advertising.
• 20:23But what if we used it for something good?
• 20:26All right. That's the intriguing possibility.
• 20:28So suddenly we've got this new kind
• 20:30of data in terms of the feasibility,
• 20:32it's objective, it's unobtrusive.
• 20:34It can be collected without bothering
• 20:36people like you do with questionnaires.
• 20:39It's individualized.
• 20:39You can get a very long baseline and an
• 20:42individual person understand their personal
• 20:44variability and deviations from it.
• 20:46It can be collected.
• 20:47It creates the possibility of real time
• 20:49responses to things that are going on.
• 20:52And of course,
• 20:53it's highly scalable because most people.
• 20:55Even in most countries around the world now.
• 20:58Have some kind of device like this.
• 21:01So that's pretty intriguing.
• 21:04So what's wrong with self report?
• 21:05This is what we've built our.
• 21:08Our whole business on in mental health and
• 21:11there's nothing wrong with self report.
• 21:13Self report is good data,
• 21:14it's interesting data,
• 21:15it's important data,
• 21:16but it's never ever complete data
• 21:19because we know that whenever we
• 21:21measure something with self report and
• 21:23with objective data at the same time,
• 21:25we find that there is either a very
• 21:27moderate correlation or sometimes
• 21:29no correlation.
• 21:31So we know that from studies such
• 21:32as there's studies on condom use,
• 21:34for example,
• 21:35that have shown this variability.
• 21:38Sleep.
• 21:38This is an area that I know a bit
• 21:40about when we use wrist actigraphy
• 21:42and we compare it to self report
• 21:44substance use where you look at.
• 21:46Say you're in screens compared
• 21:47to self report,
• 21:48you always find you get a different data
• 21:50set from the objective and the subjective,
• 21:52so having them together is really helpful.
• 21:54And like I said,
• 21:56the teenagers are pouring data into this.
• 21:58This is now quite out of date.
• 22:00You know that one of the challenges
• 22:01in this whole world is that.
• 22:042018 is a million years ago.
• 22:06In terms of the products that are
• 22:08available and how people are using them,
• 22:10but even there you can see this
• 22:12massive rise in the ownership of
• 22:15smartphone engagement with social media.
• 22:18The preference of how you like
• 22:21to communicate. So texting is
• 22:23now more popular than in person.
• 22:25Communication, social media and
• 22:27video chatting more more popular.
• 22:29So as a result of this,
• 22:30we developed a research platform called
• 22:32EARS that stands for effortless assessment
• 22:35research system, but it actually.
• 22:37There's also good reason to use the Bunny.
• 22:41And so we've got our little Bunny logo there.
• 22:43And so it's a, it's a,
• 22:44it's a research tool that people can use.
• 22:46You can download it onto your
• 22:49iOS or Android phone.
• 22:50When you when you're on board,
• 22:51you get a code from the research study,
• 22:53and that tells you that.
• 22:55Tells us that you've consented and that
• 22:57you're and tells us which study you're in.
• 22:59It emphasizes using the phone only,
• 23:01so you don't need to.
• 23:02No one needs to own a wearable
• 23:03or anything like that.
• 23:04You don't have to get them out to people,
• 23:05active graphs,
• 23:06or other weird things that
• 23:08you might want them to wear.
• 23:10They there's no special instructions,
• 23:12just use your phone as you normally
• 23:14would and so and it also is is
• 23:18collecting data continuously.
• 23:19And we can collect a lot of
• 23:20different kinds of things.
• 23:21There's, you know, there's the raw sensors.
• 23:23And then really what we've been
• 23:25working on in our research work is
• 23:27how to extract meaningful behavioural
• 23:29features from the rural sensors.
• 23:30So, for example,
• 23:31we can look at patterns of phone and
• 23:34app usage, language and cognition.
• 23:35I'll present you some data on
• 23:38that in a second.
• 23:39Not as much detail as you'd
• 23:40get from a wrist actigraph,
• 23:42but you can certainly estimate
• 23:44sleep onset and sleep offset from
• 23:46the phone usage with some accuracy.
• 23:49Had some physical activity,
• 23:51geographic location.
• 23:53We've done some work on facial
• 23:55expression and selfies put
• 23:56asterisks on that one bit hard.
• 23:58Most people use Snapchat to take
• 24:00selfies and Snapchats got its
• 24:01own camera that we can't get to,
• 24:03so in fact,
• 24:04we don't get a lot of selfies
• 24:06for that reason.
• 24:07What music people are listening to.
• 24:09Circadian patterning have some methods for
• 24:12collecting acoustic voice data as well.
• 24:14So that's a lot of stuff that you can get.
• 24:16So I want to,
• 24:17you know,
• 24:17so we could go on about this
• 24:19stuff is a great deal,
• 24:20but let me let me focus on one
• 24:22particular thing is that we,
• 24:23so we've got a method for
• 24:25collecting the language that types
• 24:27that's typed into the keyboard.
• 24:30And every keystroke is marked with
• 24:31a time and date stamp and also
• 24:34what app is in the foreground,
• 24:36so we're able to actually look
• 24:38at different language patterns
• 24:39in different apps and so forth.
• 24:41One thing that you'll notice
• 24:43here is this is a,
• 24:45this is a youth population.
• 24:46This is. There's enormous variability in
• 24:48how much people type into their phones.
• 24:50Some people type an enormous amount. This is.
• 24:53This is average daily communication,
• 24:54so we have one participant here
• 24:57who's typing in nearly 3000 words
• 24:59a day on average. On average.
• 25:01And most of them are going into social media.
• 25:04The rest are going into SMS.
• 25:06Almost nothing in e-mail.
• 25:08Right, So e-mail is definitely your
• 25:12grandfather's way of communicating.
• 25:15So I wanted to show you a little bit of
• 25:17data from one of my graduate students,
• 25:18Elizabeth Mcneely.
• 25:19This is a paper that's in press
• 25:21and clinical psych science.
• 25:23And this is with a group of
• 25:2513 and 14 year olds.
• 25:26So there are people who haven't
• 25:28had their phone for that long.
• 25:29And we we collected these messages and we
• 25:33also collected them across a period of time.
• 25:37Uh.
• 25:38And we would.
• 25:39And we what we did is we not only
• 25:41had baseline data on the sample,
• 25:42but we also had daily reports of mood.
• 25:46And what we can and we could,
• 25:48we restricted it to the the language
• 25:51that was typed into social applications,
• 25:54so either social media or messaging.
• 25:57Instant messaging.
• 25:57So wasn't search not that search,
• 25:59search date is very interesting,
• 26:01but we we did we we didn't include that here,
• 26:03but it's different.
• 26:04You've got a different audience.
• 26:06So a couple of interesting findings.
• 26:08First of all, we found that kids who
• 26:12were had lower levels of of well-being.
• 26:15I generally typed more into their phone
• 26:19and that that was particularly true
• 26:21if we looked at the daily word count.
• 26:23There's an interaction here where the
• 26:26kids who are have low well-being in
• 26:29general are particularly likely to type
• 26:31more words when they're having a bad day.
• 26:34OK, so there's an interaction
• 26:36there that's interesting.
• 26:37So overall,
• 26:38word count is important.
• 26:40First person pronouns turn out
• 26:42to be a really important marker,
• 26:44and what we're seeing here is once again,
• 26:46the kids who are more depressed in general
• 26:50have more first person pronoun use and.
• 26:53On days when you're reporting
• 26:55lower well-being you,
• 26:57you tend to use first person pronouns more.
• 26:59So what we're seeing here is the the heavy
• 27:01line is the average regression line.
• 27:03But the interesting thing to me is that
• 27:05all these others are the individual
• 27:07regression lines for each participant.
• 27:08And you can see that that slope is there
• 27:11for almost everybody in the sample.
• 27:13So it's quite consistent within
• 27:16person phenomena.
• 27:17The other interesting finding is we
• 27:19found that present focused words,
• 27:21so present tense.
• 27:22Is also more common when you're feeling.
• 27:27On when you're a person who
• 27:28doesn't feel as well in general,
• 27:30and also on days when you're
• 27:32not feeling as good.
• 27:35Now you may be interested,
• 27:36you may be wondering why why aren't we
• 27:37looking at positive and negative words?
• 27:39You know, these valence and actually
• 27:41they they don't have as strong a
• 27:43relationship as these patterns do.
• 27:45And in fact this is not,
• 27:46we're not the first people to observe this.
• 27:48This effect with first person pronouns
• 27:50has been well established and published.
• 27:53There's a meta analysis on it in the
• 27:55literature where one of the first
• 27:56to look at it in this kind of data,
• 27:58but in other language data this
• 27:59is a common factor.
• 28:00And in fact there's a there was a
• 28:03really cool paper published by Eric.
• 28:05I'm sure many of you know.
• 28:08In PNAS earlier this year and they were
• 28:10looking at therapy transcripts from
• 28:12I think it's talk. Talks talk space.
• 28:14There's a talk life in the talk space.
• 28:17I get the muddled up sometime.
• 28:18This is a an online text based
• 28:20therapy platform and they found and
• 28:22they looked at this concept of what
• 28:25they called psychological distance
• 28:27and psychological distance was
• 28:28the inverse of using first person
• 28:31pronouns and present tense words.
• 28:33So in other words,
• 28:34they saw it as a good thing.
• 28:35They saw it as a capacity
• 28:37to decentre or to regulate.
• 28:39And sure enough, they saw that.
• 28:43There was an effect across treatment.
• 28:47That symptoms got better.
• 28:50Linguistic distance defined as I
• 28:54just described. Improved, improved.
• 28:56And that there was a relationship between
• 28:59this linguistic distance and symptoms,
• 29:02both between subjects and within subjects.
• 29:06So it's cool paper,
• 29:07definitely worth checking out.
• 29:08So that's a kind of a replica,
• 29:09you know,
• 29:10a consistent finding to the one I've
• 29:12just presented you from Elizabeth's work.
• 29:14So the interesting thing about this is that.
• 29:20One of the unique affordances of this kind
• 29:23of approach is the ability to actually look
• 29:26at day-to-day variability within a person.
• 29:28And this turns out to be something
• 29:31that's quite of interest to clinicians.
• 29:33This is from a blog that they published on.
• 29:38Google X, which is their research arm and
• 29:41what they were doing this work on a on an
• 29:44EEG system that would diagnose depression.
• 29:47And as part of that they went out
• 29:48and they talked to clinicians,
• 29:49they did what they called in
• 29:51business customer discovery.
• 29:52So they went and talked to the
• 29:53clinicians and they said, you know,
• 29:54would you like to have this EEG system
• 29:56that could help diagnose depression.
• 29:57And interesting thing was most
• 29:59of them said not really.
• 30:00Because I actually, I'm pretty,
• 30:02I think I'm pretty good
• 30:03at diagnosing depression.
• 30:04You know, I've been doing it for a long time.
• 30:06I've been trained well to do it,
• 30:07you know, and it's it's it's, you know,
• 30:10probably in some ways not that hard,
• 30:12you know, to diagnose depression.
• 30:14And so they said,
• 30:15but what we would really love and this
• 30:17is where we really have a problem, is.
• 30:19Using technology as a tool
• 30:21for ongoing monitoring,
• 30:22knowing what's working and not working,
• 30:24not what's not working.
• 30:25How is my patient doing now?
• 30:28How are they?
• 30:28How is their trajectory of what's going on?
• 30:30Do I need to change my therapeutic strategy?
• 30:33You know,
• 30:33things like that is actually where
• 30:35the clinicians were much more
• 30:36interested than that first thing.
• 30:37And so we've got a technique here that at
• 30:40least potentially can contribute to that.
• 30:42So let's go back to our grand challenges.
• 30:46Prevention, access. Quality, effectiveness.
• 30:50And the fact is, if you read most things
• 30:53that people write about digital technology,
• 30:56they're going to say, why is it exciting
• 30:58in mental health services as well?
• 31:00Because we can address the access problem.
• 31:03All right, so we can give people
• 31:04access an app, you can download it,
• 31:06everyone can have access.
• 31:07We can send it all around the world.
• 31:09Software solution,
• 31:09take the human out of the loop. Brilliant.
• 31:14There's a problem. First of all.
• 31:17We know that increasing access.
• 31:20In and of itself does not help.
• 31:24And how do we know that?
• 31:25Because in high income countries like the US,
• 31:28the UK, Australia and Canada,
• 31:31there has been a massive increase in access
• 31:33to mental health services over the last
• 31:35few decades with no concurrent reduction.
• 31:39In the burden of disease associated with.
• 31:43Mental health problems,
• 31:44so we're pouring more money in.
• 31:47It's a good thing I've supported.
• 31:50But we're not getting a better outcome.
• 31:52So access itself is not a solution
• 31:54and when you come to digital products,
• 31:56there's a very clear reason for that
• 31:59is that is that they're not sticky.
• 32:01So this is a curve of how many
• 32:03people keep using an app.
• 32:04This is mental health apps
• 32:05specifically over a period of time,
• 32:07and you can see within three to four days,
• 32:1080% of your people you've lost.
• 32:12So this is The Dirty little
• 32:14secret of digital health, right?
• 32:15The fact is that you're actually,
• 32:17it's what David Moore has.
• 32:20Brilliant on this issue calls
• 32:22the denominator problem.
• 32:24You know,
• 32:24when you run an RCT with a digital product,
• 32:26you can show effectiveness,
• 32:28but often they don't tell us
• 32:30how many people wouldn't use it.
• 32:33And so that's a big problem.
• 32:35We do see a little signal over here.
• 32:38See this Gray line?
• 32:39These are the apps that have a
• 32:41social element. Oh my goodness.
• 32:45Someone's gonna go away.
• 32:50Umm. Yeah. And anyway,
• 32:52so we'll move along and so this and so we.
• 32:55And so one of the themes that's coming
• 32:57through with digital technology and
• 32:59it's usefulness and mental health is
• 33:01that having a human in the loop in
• 33:04some way seems to be really important.
• 33:07It seems to make a difference
• 33:08to effectiveness and engagement
• 33:10and of course, effectiveness,
• 33:11engagement are related to each other.
• 33:14So we can't, you know,
• 33:15so I think we need to reframe the problem
• 33:17from one where we're saying like,
• 33:19how can we just scale things
• 33:21and get the humans out.
• 33:24To say like,
• 33:25let's try and understand what the humans are
• 33:27trying to do and help them to do it better.
• 33:30And that's, I think,
• 33:31a better version of how we can
• 33:33move forward with digital.
• 33:35Hello, I've lost my.
• 33:39Kinko's got rid of my control of the screen.
• 33:41OK, so back to our problems.
• 33:44So we really do need to solve
• 33:45the quality and effectiveness
• 33:46problem along with the access
• 33:48problem or we're not doing,
• 33:50we're not having much impact.
• 33:54And, you know, one of the
• 33:56interesting things is that we. Umm.
• 33:59This is term they use in digital
• 34:02technology called scudo morphism
• 34:04is anyone heard that skeuomorphism
• 34:06is what Apple products used to
• 34:08do when they when they had,
• 34:10when they gave you a calendar,
• 34:11the little icon for calendar looked
• 34:13like a literal calendar, right.
• 34:15And so that's that's that's.
• 34:17And so in some ways,
• 34:19one of the problems we have with digital is
• 34:21that we're we're using skewed amorphism.
• 34:23This is not my idea.
• 34:24This is Stephen Schuler,
• 34:25University of California,
• 34:26Irvine made this point in a brilliant paper.
• 34:29We're we're saying like what we
• 34:31need to do with digital is make
• 34:34it like a clinical consultation,
• 34:35like make it like the way we
• 34:38currently do things.
• 34:39But the but the thing about that
• 34:41is that actually the way we do
• 34:43things now actually might be
• 34:45quite poorly matched to the actual
• 34:47thing we're trying to do.
• 34:48Let me explain what I make that
• 34:50didn't come out very clear at all.
• 34:51Let me explain what I'm trying to mean.
• 34:53We know.
• 34:55Pretty well what the principles
• 34:58of behavior change are.
• 35:00It's something like this.
• 35:01You start with a clear description
• 35:03of the new skills to be learned,
• 35:05along with modeling of those skills so
• 35:07present to know what they want to do.
• 35:09Then they've got to have chances
• 35:11for practice,
• 35:12repeated practice with timely
• 35:13feedback so that they can actually
• 35:16keep trying and failing,
• 35:17but getting good feedback,
• 35:18and so that they learn.
• 35:20And then you have to have specific
• 35:22procedures to make sure that
• 35:23the skill generalizes to the
• 35:25environment you care about.
• 35:26So it doesn't matter whether it's throwing
• 35:28a football or driving a car or whatever.
• 35:30The skill is.
• 35:31The formula is something like this,
• 35:32and the things that we're
• 35:34actually good at teaching people,
• 35:35we tend to follow this formula.
• 35:37Now I put it to you that the current way.
• 35:40We do psychotherapy is a little bit
• 35:43like a football coach who says?
• 35:46Listen, I'm not going to come to the game.
• 35:49And I'm not even gonna watch
• 35:51tape of the game.
• 35:52What I'm going to do is I'm going
• 35:53to sit in my office and I'm going
• 35:54to have you come and tell me once
• 35:56a week how you think you played.
• 35:57And then I'll give you some verbal
• 35:59suggestions about how you could
• 36:01play better and then you can
• 36:02go out and play again and then
• 36:04come back next week and repeat,
• 36:06repeat the process, right.
• 36:07So that's.
• 36:08So in that sense,
• 36:09what I'm trying to point out is
• 36:11the way we deliver services now
• 36:13in these kind of appointments,
• 36:15punctuated appointments is actually
• 36:16very poorly matched with the process of.
• 36:19Behaviour change with what we know
• 36:21about the fundamentals of behaviour change.
• 36:24So this is where I think digital
• 36:26technology has an affordance to actually
• 36:29think about doing things differently.
• 36:31Because we can take that therapy
• 36:33process out of the office,
• 36:35into People's Daily lives,
• 36:36and it can be portable and it can be in
• 36:39your pocket and all that kind of thing.
• 36:41So we've been working on a long after years
• 36:45which is still being used by many people,
• 36:48we've developed a new system called Vera.
• 36:52And what this is,
• 36:54is it's a system that's designed to bring
• 36:56this capacity for continuous remote
• 36:59patient monitoring in behavioral health.
• 37:01And the capacity for ongoing
• 37:04continuous behaviour change support
• 37:06as a tool set to the clinician
• 37:09and so it goes kind of like this.
• 37:11You have this patient app.
• 37:15The person downloads and puts
• 37:16it on their phone.
• 37:17It collects data continuously.
• 37:19If the person wishes to,
• 37:22they can then share those data
• 37:24with a practitioner who has the
• 37:26complementary so well they only software.
• 37:28It's a web portal,
• 37:30so they need to log in and then the
• 37:33practitioner and what's happening is.
• 37:35What we're doing is we are collecting
• 37:38data on mood and behaviour across
• 37:41time within an individual and
• 37:43then we're building a model.
• 37:45A data model of each individual.
• 37:48And what are the particular
• 37:50behaviour patterns that support
• 37:51good mood and well-being for them?
• 37:53So this is simple.
• 37:54This is very simple stuff folks.
• 37:56This is called behavioral activation, right?
• 37:58It's an old technique, but one with a great.
• 38:02Pedigree in terms of effectiveness,
• 38:04but when I was a lad and we were learning,
• 38:07what would we do? We give people these.
• 38:11Matrix,
• 38:11matrix questionnaires.
• 38:11And they'd have to fill in what they
• 38:13did every hour and how they enjoyed it
• 38:15and whether it gave them a sense of mastery,
• 38:17as it was called then and so forth.
• 38:19And and and of course people didn't do it,
• 38:21you know,
• 38:22or they'd fill it in in the waiting
• 38:23room before they came to see you.
• 38:25All that kind of stuff, right?
• 38:26So this takes the effort out of it,
• 38:28because we're actually measuring behavior
• 38:30continuously and and objectively,
• 38:32and we're also correlating it
• 38:34with mood variability.
• 38:35And then for each interpersonal we
• 38:36are able to tell you which aspect of
• 38:39mood variability is most important.
• 38:42You. Then.
• 38:43That is fed back to the practitioner.
• 38:46The practitioner can look at
• 38:48those suggestions.
• 38:48She's not getting much sleep.
• 38:51Umm, you know,
• 38:52you can look at patterns like rigid
• 38:54thinking and the language and so forth.
• 38:56And it will also make some suggestions like,
• 38:58well, look,
• 38:59when this person gets more sleep,
• 39:00their mood is better or when they're more,
• 39:02when they have more social connection
• 39:05or if they are more positive in their
• 39:08thinking style or if they are more
• 39:11physically active or whatever it might be.
• 39:13And so that's fed back to the clinician.
• 39:15The clinician then build a,
• 39:17an intervention that is supported
• 39:20by just in time nudges.
• 39:22So they actually build out.
• 39:24They discussed this with the
• 39:25person and of course you know
• 39:26the idea of nudging this term.
• 39:27It comes from behavioral economics.
• 39:29The important thing about a
• 39:30nudge is that the person's got
• 39:32to match a goal the person holds.
• 39:33You can't manipulate people's
• 39:35behavior with a nudge,
• 39:36but if someone's trying to do something,
• 39:37then the nudge might connect their
• 39:40current self with their future
• 39:41self for one of a better term.
• 39:43You know, connections to that
• 39:44long term goal that they have.
• 39:46And so we know that nudging is not perfect,
• 39:48but it does.
• 39:48You know, the behavioral economics
• 39:50literature suggests that, you know,
• 39:52in all these different areas where
• 39:54they've looked at vaccines and, you know,
• 39:55various things, eating behavior,
• 39:56there's a lot of studies.
• 39:58Meta analysis tend to suggest it works.
• 40:00It's good.
• 40:01It helps people more often,
• 40:02more of the time follow through
• 40:04with their behavioural intentions.
• 40:06So you can build this in and this then
• 40:08arrives on the phone as a notification.
• 40:10Person can click through on that,
• 40:13they can let us know if they did it or not.
• 40:17And they can then and then we can learn.
• 40:21And the other thing that and the
• 40:23thing that actually gets me very
• 40:25excited about this is that what we're
• 40:27also doing is we are digitizing
• 40:29all this stuff in the workflow.
• 40:31Right.
• 40:31So this is so the clinician doesn't
• 40:34have to spend more time on.
• 40:35This is something I use when
• 40:37they're with the patient.
• 40:37And so suddenly we can then roll that up to
• 40:40the healthcare organization and they can
• 40:43see in real time are people getting better?
• 40:46What are my clinicians doing?
• 40:48Because at least in most
• 40:50evidence based therapies,
• 40:50some kind of homeworkers
• 40:52usually part of the deal.
• 40:55You see which kind of you can see what's
• 40:57working with our population here.
• 40:58Not some RCT across the other side of the
• 41:01country or the other side of the world,
• 41:03but with my population,
• 41:04with its particular characteristics,
• 41:06what is working and so that I
• 41:08can use that for supervision,
• 41:10for training, for service reform,
• 41:12all that kind of stuff.
• 41:14So suddenly we're creating a learning system,
• 41:17and one of the things that I think is
• 41:19exciting about this is the fact that
• 41:21we actually know more about traffic.
• 41:22Than we do about mental health.
• 41:25Because we get real time data and
• 41:28we adjust policy and interventions
• 41:30on the basis of it.
• 41:32And we don't get that in mental health.
• 41:34So that's that's so that's what we're
• 41:36working on and we're we're currently
• 41:37where we are with that is it's built,
• 41:40it's working and it's needs to work better,
• 41:42but it's but we're working on it.
• 41:45We've got a series of partnerships
• 41:47with different groups to test this out
• 41:49with different kinds of populations
• 41:51and and certainly if anyone here is
• 41:53interested in talking about this,
• 41:55I'd love love to chat to you.
• 41:58Umm.
• 42:00So. So what are the next steps?
• 42:04Well, the next steps are.
• 42:08At the moment we decided to start with
• 42:10a system where this kind of nudging
• 42:12procedure is controlled by clinician.
• 42:14But there are, and you know,
• 42:16there are various reasons for that.
• 42:17We want to learn more about it.
• 42:18We want to collect data.
• 42:19We want safety.
• 42:21Etcetera, but the.
• 42:23There's a lot of people talking about
• 42:25this concept of just in time adaptive
• 42:28interventions and so these are more
• 42:30automated things that are driven off.
• 42:33Machine learning,
• 42:34artificial intelligence, say.
• 42:35Just like that.
• 42:36But here's a definition,
• 42:37an intervention design that aims to provide
• 42:39just in time support by adapting to the
• 42:41dynamics that have been individuals,
• 42:43internal state, and context,
• 42:45which is measured continuously.
• 42:47Italics are mine. So.
• 42:51The point is that with mobile sensing.
• 42:54There's a lot of problems.
• 42:55Let me be clear.
• 42:56In case I'm sending to evangelical,
• 42:58I want to be clear.
• 42:59There's a lot of unresolved
• 43:01problems with mobile sensing.
• 43:02All right,
• 43:03we've got a lot of work to do to
• 43:05get better at it and to understand
• 43:06more what the data means.
• 43:08But it but it but at least it
• 43:10gives us line of sight on that.
• 43:13Problem.
• 43:13So let me give you a very,
• 43:15very simple example,
• 43:16one that I'm excited to be
• 43:18working with some colleagues on
• 43:19who are interested in people
• 43:21who are addicted to opiates.
• 43:23We're thinking about a procedure where.
• 43:26Participant could.
• 43:28Identify locations.
• 43:30Dropping a pin on a map that they
• 43:33are now a danger locations for them.
• 43:35So it could be the doctor where
• 43:37you get prescribed your opiates.
• 43:38It could be the,
• 43:39you know where you score your heroin.
• 43:41It could be, you know, like,
• 43:42so it could be wherever.
• 43:43It could be your friend who you go
• 43:44and visit and you used together,
• 43:45you know.
• 43:46So there's various things and you
• 43:47could drop those in if you wish to.
• 43:48This is all you know,
• 43:50totally only if the person wants to do it.
• 43:52And then it's not a hard technical
• 43:55problem to then geographically ring fence
• 43:57those locations and give the person a nudge.
• 44:00Whenever they're nearby.
• 44:03They're still free to do
• 44:04whatever they want to do.
• 44:04We're not controlling their behaviour,
• 44:06but we're just reminding them that if
• 44:08they do have a commitment to sobriety.
• 44:10That this is a decision point for them.
• 44:13And that we're raising that awareness,
• 44:14like I say,
• 44:15connecting your future self
• 44:17with your current self.
• 44:18So that's the kind of thing
• 44:21that's actually very tractable.
• 44:23The other one that I haven't
• 44:24talked about a lot today,
• 44:25but is that is really this was
• 44:26the use case that got me excited
• 44:28in the 1st place about it,
• 44:29which was suicide prevention.
• 44:32This is a harder problem.
• 44:34But we know that all the interventions
• 44:37that we have that are effective for
• 44:39suicide prevention work because
• 44:41they have an impact at the time of
• 44:43high risk and the time of high risk
• 44:45where there's an intention to act,
• 44:46and the time of high risk for
• 44:48most people is relatively short.
• 44:50Might even be only 10 or 15 minutes long,
• 44:52but the things that work are things
• 44:54that put some friction in at that point.
• 44:57Don't have a gun in the house?
• 44:59Keep your ammunition away from the gun.
• 45:02Put pills into blister
• 45:03packs instead of bottles.
• 45:05Put bridge, barriers up, etcetera.
• 45:08Don't, don't give people access to
• 45:10agricultural chemicals that have,
• 45:11you know, they're highly toxic.
• 45:12There's a whole range range of
• 45:14these things that have all been
• 45:15shown to be effective at the point
• 45:17is they all have an impact on at
• 45:18the at the moment of high risk.
• 45:20So it's a harder problem than
• 45:22the one I just described for
• 45:23the substance use scenario.
• 45:25But if we could get even some
• 45:27probabilistic estimate of when
• 45:29people's mood if someone's at risk.
• 45:31And then their mood is shifting
• 45:33in the particular direction,
• 45:34then the capacity to reach out at
• 45:36that time might be something that
• 45:37could really be a game changer.
• 45:39So that's pretty exciting. The.
• 45:43So the ultimate vision for this,
• 45:44also in terms of access,
• 45:46is to actually build a system that provides.
• 45:51Access to what people call stratified care,
• 45:54right.
• 45:54So one of the problems we have with
• 45:56our current mental health system is
• 45:58that we use the same solution for
• 46:00almost every problem most of the time.
• 46:02So it's expensive.
• 46:03People like me, you know,
• 46:04and and you know,
• 46:06it's the 50 minute hour and you know,
• 46:07we have that kind of,
• 46:08it's not always true,
• 46:09but but there's a tendency to
• 46:11use that one solution and what
• 46:13we want to be able to do is.
• 46:20Give people, through a digital tool, access
• 46:23to a range of different levels of care.
• 46:27That they can access. Without friction.
• 46:32Mental healthcare systems full of friction.
• 46:34It's hard to get in,
• 46:35hard to know where to go,
• 46:36hard to get an appointment,
• 46:38hard to know if you're not doing well
• 46:40because it's your fault or the therapist.
• 46:41You know? There's just like
• 46:43fiction everywhere and so.
• 46:45Once you're in,
• 46:46you can use the app for self-care.
• 46:48You can interact with automated nudges.
• 46:50You can get a text based coaching model,
• 46:53health coaching model,
• 46:54which is more scalable than a
• 46:56traditional face to face therapy model.
• 46:57You can also access telehealth.
• 47:00Consultations and face to
• 47:01face therapy and if necessary,
• 47:03hospitalization and inpatient care.
• 47:07A daycare, sorry.
• 47:10But the idea is,
• 47:11wouldn't it be cool if someone
• 47:12who had a mental health challenge
• 47:14could plug into this system at
• 47:15the right level for them and then
• 47:17move up and down according to what
• 47:18they needed at a particular time,
• 47:20knowing that they can still move back
• 47:23up or move back down relatively easily?
• 47:26So that's that's the kind of the goal.
• 47:28So we're starting on.
• 47:30This part and this part and this part,
• 47:34but we want to kind of build it out overtime.
• 47:37All right.
• 47:38Thank you for listening.
• 47:54Thank you very much, Nick.
• 47:55And questions for Doctor Allen.
• 48:02Carla.
• 48:07So I'm very interested in this.
• 48:09I do a lot of work with people
• 48:11with significant trauma
• 48:12backgrounds and a lot of emotion.
• 48:13Emotion dysregulation,
• 48:15difficulties, irritable,
• 48:16angry, outbursts, violence.
• 48:18And I'm thinking about this
• 48:20in terms of applications to
• 48:22someone being able to monitor
• 48:24who isn't very well in touch
• 48:26with their internal States and
• 48:27being able to monitor them
• 48:28and get pings that they need to use
• 48:30coping strategies. Does that seem
• 48:32like something that is?
• 48:35A million years away,
• 48:37or potentially feasible sooner? So.
• 48:40Well, let me go back to the microphone.
• 48:47So when I did my PhD research,
• 48:49I did it in psychophysiology and
• 48:52so this is not a new problem,
• 48:54this is an old problem and I was particularly
• 48:56in the psychophysiology of emotion.
• 48:58And so one of the problems we know that
• 49:00we've got with the psychophysiology
• 49:02or behavioural signatures of
• 49:03emotion is that there's, it's a,
• 49:05it's a kind of a many to one problem.
• 49:07There's lots of different.
• 49:09Signatures that can occur that
• 49:11can be associated with emotional
• 49:13experience across individuals.
• 49:15So one person might show high autonomic
• 49:17arousal and another person might not,
• 49:19and another person might express
• 49:20it in language, you know,
• 49:21so there's lots of so.
• 49:23So I want to say that as a background
• 49:25that that there's a lot of people
• 49:26in this digital health area and you,
• 49:28if you read the digital tech press and
• 49:30the saving some of the mainstream press,
• 49:32you'll see these things occasionally.
• 49:34Does your smartphone know when you're
• 49:36depressed, you know, or something like that?
• 49:37And it's always simple.
• 49:39Have simplified.
• 49:41Because of this many to one problem
• 49:44this multifinality issue but.
• 49:46There is a,
• 49:47I think a tractable version,
• 49:49which is that the system has to
• 49:51learn about the individual right.
• 49:53It has to learn at the individual level.
• 49:55And I think that's more tractable.
• 49:56If you could actually track someone
• 49:58across time and understand their
• 50:00mood variability and then look
• 50:01at the particular signatures that
• 50:03are associated with them,
• 50:04that's what we do as clinicians, right?
• 50:07We try to learn what for each
• 50:09individual intervention, you know,
• 50:10if they're working with the clinician,
• 50:11trying to identify what are their signals,
• 50:13then you could input it into
• 50:15the device exactly.
• 50:16So for example.
• 50:17You know, there's lots of examples.
• 50:19The one that you mentioned,
• 50:20another classic is someone who
• 50:22experiences bipolar disorder.
• 50:24And you know,
• 50:24when you work psycho therapeutically
• 50:26with someone with bipolar disorder,
• 50:27one of the main things you do is
• 50:29try to help them to learn early,
• 50:30particularly in the manic phase to pick up.
• 50:33That's something shifting,
• 50:34which they often don't notice until too late.
• 50:38But over repeated episodes
• 50:39they can start to learn.
• 50:40And so if you could have something like
• 50:42that that is monitored and fed back,
• 50:44then that's very empowering
• 50:46to the individual and I.
• 50:47And I just, I'll dwell on that point,
• 50:50empowering for a moment.
• 50:51Because one of the things I
• 50:52suspect some of you are thinking,
• 50:53if not all of you is.
• 50:55It's a common reaction.
• 50:56I get people go, this is a little creepy,
• 50:58nick, this is a little creepy.
• 51:00You know,
• 51:01you're tracking people and things like that.
• 51:03And I think there's a couple of,
• 51:05there's good reasons for that, but
• 51:07I think one thing that's important is that.
• 51:10Data is power.
• 51:13And the question is,
• 51:14who is it empowering?
• 51:16Now,
• 51:16our most common experience with
• 51:18digital data at the moment is that.
• 51:20It empowers large tech companies who
• 51:23use it to manipulate what we buy.
• 51:27But there is actually a use of data,
• 51:29I believe, that is empowering to the
• 51:31individual if you set it upright.
• 51:33And this is the kind of use
• 51:34case you're talking about.
• 51:36Like, it's empowering to me to understand.
• 51:39My patterns of behaviour and how they
• 51:41are going to affect my long-term health.
• 51:45It's empowering to me to be able
• 51:47to share those data with someone
• 51:48I want to share it with,
• 51:50as well as take the data back when I don't
• 51:52want them to have it anymore, right?
• 51:54Which our system does.
• 51:55So I do think.
• 51:58Yes, it's potentially creepy and you've gotta
• 52:00and you know you've gotta manage it, right?
• 52:02But I think if you if we ignore it,
• 52:05then we're missing out on an
• 52:07opportunity to empower people with
• 52:09greater knowledge and self-awareness.
• 52:10Yeah.
• 52:15Susie?
• 52:21Thank you so much for this presentation.
• 52:22I have multiple questions,
• 52:24but I'm going to go with one and that is
• 52:27I'm given the deluge of funding that is
• 52:29coming out of the federal government,
• 52:31whether the bipartisan infrastructure
• 52:33law or the safety bill,
• 52:37there's huge resources for digital work.
• 52:41The question is whether you are interfacing
• 52:44with policy to show this additional,
• 52:46this additional value,
• 52:48this hugely important.
• 52:49Value that is offered through this.
• 52:51And so are you doing it or are you
• 52:53interested in playing a policy role?
• 52:55So when you say you,
• 52:58you mean me as an individual,
• 52:59I mean you as an individual,
• 53:01University of Oregon, your private company.
• 53:03The time is now,
• 53:04the money's out and and and I
• 53:05think that there would be a lot of
• 53:07interest in what you're doing because
• 53:09you're showing an additional value
• 53:11to bridging that digital divide.
• 53:13So the quick answer to that is.
• 53:16I'm trying,
• 53:16but it's not something I'm necessarily
• 53:19well trained in or good at.
• 53:21So, for example,
• 53:22I mentioned briefly this report that we
• 53:25did for the world Innovation Health Forum.
• 53:28I've just finished working on a white
• 53:30paper with the National Scientific
• 53:32Council on the developing adolescent,
• 53:34which is a a group I'm involved in
• 53:37which is trying to influence policy.
• 53:41I'm trying to think of some other examples.
• 53:43You know I've been to meetings with
• 53:44UNICEF and the World Economic Foundation
• 53:46and other sort of policy related things.
• 53:48So trying, but I will say I'm a
• 53:52clinical psychologist, you know,
• 53:53I like seeing patients and collecting
• 53:55data and building software now, but I,
• 53:57you know, it's it's an interesting skill set.
• 53:59I don't know that I'm, I'm,
• 54:01I need people to help me.
• 54:04In the water, yeah.
• 54:05So now if you start swimming, yeah.
• 54:07No, but I think you're absolutely right.
• 54:09And of course, as you know,
• 54:11whenever there is an A deluge,
• 54:13as you describe it,
• 54:14of government funding, there's always,
• 54:16even when you like the topic they're funding,
• 54:18there's always that feeling like.
• 54:20How much of it is going to get wasted?
• 54:22And we know it's some,
• 54:25but how can we minimize the
• 54:26amount that's going to get wasted?
• 54:27And there's, of course,
• 54:28the moment the government,
• 54:29there's a bucket of money
• 54:30from the government,
• 54:31all sorts of hucksters and so forth
• 54:32will be running at it along with the
• 54:34people who've actually got solutions.
• 54:35So I think partly the digital area has
• 54:40been very fatty and and influenced by that.
• 54:43I think the good news is that what I'm
• 54:45seeing even in the business part of it
• 54:47and Inventure venture funding and so forth,
• 54:49there's a,
• 54:49there's a maturing.
• 54:50Where people are starting to ask
• 54:53better questions of people who've
• 54:54got some new app and they're asking
• 54:57for evidence and they're asking for.
• 54:59You know,
• 54:59implementation feasibility and
• 55:01all that kind of stuff.
• 55:02So I think we're getting better,
• 55:03but there has you know it,
• 55:05it there's definitely a potential for.
• 55:08Poor spending.
• 55:10Angie has a question online.
• 55:14Thank you. Thank you so much for this talk.
• 55:17It actually is very refreshing to
• 55:20actually have data behind things that
• 55:23you clinically see in the day-to-day.
• 55:26So a lot of the research that we do
• 55:30is suicidality and adolescence and
• 55:33so you know just anecdotally things
• 55:35that you know that is that your your
• 55:38patients listen to more dark music.
• 55:41You know, is there a way,
• 55:42is there a way with which we could
• 55:44track their Spotify and see, you know,
• 55:47kind of what they're listening to so that
• 55:49we can then send them a nudge and say,
• 55:51hey, you know,
• 55:52you're listening to too much of that
• 55:55music or whatever it is that they're.
• 55:58But then one of the things that
• 56:01we've come across is from,
• 56:04from the up down perspective of healthcare.
• 56:10Jokingly said e-mail is the your
• 56:13grandfathers but but right now that is
• 56:16how we're allowed to communicate with
• 56:19our patients when clinically we know.
• 56:22Just from just from our day-to-day
• 56:24interactions I know that I'm better
• 56:27able to reach out to my my participant,
• 56:30my adolescent participant.
• 56:31If I text them, more likely to text me back.
• 56:37From what I do in research,
• 56:39then applying that to outpatient services,
• 56:43there's a big divide because
• 56:45obviously HIPAA distractions.
• 56:47You know how how do we protect
• 56:50the patients information?
• 56:52How does this interface all
• 56:53of the data that is collected?
• 56:55How does that interface with
• 56:57EPIC for example as a, as an ER?
• 57:01So I'm I'm wondering.
• 57:04How are you or how do you envision
• 57:09taking some of the the data that you've
• 57:12acquired to convince some of these
• 57:16larger echelons of healthcare to allow
• 57:19for us to have a better relationship,
• 57:22even if it is through these technological
• 57:26devices with our adolescents,
• 57:28because developmentally that's
• 57:29where they're at and that's
• 57:31what they're expecting from us.
• 57:33We need to be mobile.
• 57:35For for our patients.
• 57:37Yeah, great. 2 great points.
• 57:40I'll take the second one first.
• 57:41So the. Yeah, you're absolutely right.
• 57:44I don't know if you've ever called
• 57:45one of your adolescent kids on the
• 57:46phone and they picked up and said,
• 57:48why didn't you text me?
• 57:49You know, because they really like
• 57:51interacting via text. And so,
• 57:53but this is actually a solvable problem.
• 57:55I mean, our system is fully HIPAA compliant,
• 57:57so you can build it to a
• 57:59HIPAA compliant standard.
• 58:00And the problem is that and and
• 58:02so then it means, you know,
• 58:03but a lot of the solutions that the
• 58:05that the adolescent is using for their
• 58:07regular texting or instant messaging.
• 58:09Or not.
• 58:10And so that's one of the advantages of
• 58:12having a kind of an app like ours is
• 58:14that you've actually got a HIPAA compliant.
• 58:17Messaging system within it and that
• 58:18the apps already on their phone.
• 58:20So there are there are ways to deal
• 58:22with that, but I agree and and and.
• 58:25Patients love it.
• 58:26I love having that kind of capacity
• 58:28to check in and it's a little bit
• 58:31like when Marshall Lanahan started
• 58:32you know having a therapist and
• 58:34DBT carried phones with them and
• 58:36everyone said like Oh my God,
• 58:37you can't do that.
• 58:38These, you know these clients
• 58:39will be just constantly ringing.
• 58:40You turns out not to be true
• 58:42most of the time.
• 58:43You know most clients actually use it very
• 58:46responsibly and so and that's even with the,
• 58:48you know, the borderline clients
• 58:49that they were working with,
• 58:50with the PT.
• 58:51So.
• 58:51So I think that you know,
• 58:53patients really appreciate having greater
• 58:55access and they and they tend to use it.
• 58:58Very um, but but you know.
• 59:01Legal systems can be built to be
• 59:03HIPAA compliant and so I think it's a
• 59:04question of how you how you do that.
• 59:05Now just remind me what
• 59:06your first question was.
• 59:07I thought it was really interesting and.
• 59:09Wanted to say
• 59:10it was about checking their Spotify.
• 59:14Really? Yeah, we can totally do that.
• 59:15So we can look at the music
• 59:18and there's actually Spotify.
• 59:20We actually use their engine because
• 59:21they have an engine for music
• 59:23recommendation and they have every song
• 59:25in the world listed and they rated on
• 59:27these musical and emotional qualities.
• 59:28But once again, the important thing is
• 59:30not all kids who listen to Norwegian
• 59:33death metal are about to kill themselves,
• 59:35right? It it it.
• 59:36But if you've got a kid who who
• 59:38doesn't normally listen to that,
• 59:40so it's about the within person change,
• 59:42right?
• 59:43And if they suddenly are listening to it?
• 59:45A lot that might be,
• 59:46I mean this is a speculation at this point,
• 59:49but so it's important not to say it's,
• 59:52I think it's more state like than
• 59:53trade like you have to understand
• 59:55the usual pattern of listening.
• 59:56We don't want to say everyone who
• 59:58loves ex kind of music is you know,
• 60:00got a mental health issue or in need.
• 01:00:02But I think if you see changes,
• 01:00:03dynamic changes across time.
• 01:00:05We do know anecdotally in clinical
• 01:00:06circumstances where sometimes kids
• 01:00:08who are getting very are having
• 01:00:10a lot of suicidal ideation will
• 01:00:11listen to a certain kind of music.
• 01:00:13That for them is very much associated
• 01:00:15with that mood state and if there's
• 01:00:17a lot of listening to that,
• 01:00:18that could be a marker.
• 01:00:20Yeah.
• 01:00:22So we are just about at time,
• 01:00:24but I would just like
• 01:00:25to thank Doctor Allen for
• 01:00:27a wonderful presentation.