Justin Baker 2

Transcript

• 00:09Hello and welcome to the Science et
• 00:11al podcast about everything science
• 00:13sponsored by the Yale School of Medicine.
• 00:15I'm your host, Daniel Barron,
• 00:17and in this episode I'm speaking
• 00:19with Doctor Justin Baker.
• 00:21Justin is the Co founding scientific
• 00:23director of the McClain Institute
• 00:24for Technology in Psychiatry and
• 00:26he also directs the Laboratory
• 00:28for Functional nor Image Ingane
• 00:30by Informatics at McLean Hospital.
• 00:32He is an assistant professor of
• 00:34psychiatry at Harvard Medical School
• 00:36and in all of these capacities he has
• 00:39the time to do research when it tries
• 00:41to combine his expertise in Bremen, Jane.
• 00:44With his expertise in deep
• 00:45multi level phenotyping,
• 00:46something that will discuss in the podcast.
• 00:50He's a clinical psychiatrist
• 00:52with expertise in schizophrenia,
• 00:54bipolar spectrum disorders and
• 00:56other disorders.
• 00:58Ann, I first learned to Justin and
• 00:59his work through a colleague at NYU.
• 01:01You the beginning of my residency training.
• 01:03It was like the very beginning
• 01:05of my intern year.
• 01:06So I read an article and actually
• 01:09written an article for Scientific
• 01:10American about this nascent
• 01:12field of digital diagnostics,
• 01:14something I thought was really cool,
• 01:16but I didn't know much about yet an
• 01:19my NYU you friend told me to check
• 01:22out Justin's research and at the time
• 01:25Justin was using digital devices like
• 01:27smartwatches or Fitbits to monitor
• 01:29and trace patients symptoms and try
• 01:32to combine that with biological
• 01:34measures like brain imaging.
• 01:35I invited myself to Justin's annual
• 01:38technology in Psychiatry Summit in Boston.
• 01:40Which was really cool.
• 01:41He had adjusted and been able to invite
• 01:44out speakers from Apple and Google,
• 01:46and he had Tom Insel give.
• 01:48What are the key notes is really
• 01:50exciting and later that winter I was
• 01:52invited to give a talk in McLean.
• 01:55Anne asked to meet with Justin
• 01:57during the day.
• 01:58I remember us walking
• 01:59around at Mcleans campus,
• 02:00which even in the dead of Winter
• 02:02was still quite lovely and he
• 02:04and I just kind of patrolled the
• 02:07perimeter and tell her fingers got
• 02:09cold and we had to go inside.
• 02:11And during this time I really got
• 02:13to know Justin and he was able
• 02:15to give me a lot of really useful
• 02:17advice to get through residency,
• 02:19and since that time's I found him to be
• 02:21a very kind and generous mentor and friend.
• 02:25I'm really grateful to Justin for
• 02:28participating in this podcast and.
• 02:30Also,
• 02:30and and all the help that he's given
• 02:32me over the last few months when I was
• 02:35writing a book about digital psychiatry,
• 02:37which Justin's a big big expert.
• 02:40This podcast was filmed at the end
• 02:42of a very busy day for Justin.
• 02:45I had invited him out to yell to
• 02:47give this psychiatry grand rounds
• 02:49and really enjoyed watching the
• 02:52audience and seeing how impressed
• 02:54and kind of awakened they seem.
• 02:56Looking at all this research that
• 02:58Justin was doing and how he could
• 03:02use these digital devices to create
• 03:04clinically useful tools and so.
• 03:06Really excited to present this
• 03:08episode with Justin
• 03:20What was your residency program like? Like?
• 03:23What was your experience there like?
• 03:26Were they? It sounds like they were trying
• 03:30to expose you to different researchers,
• 03:32and I'm curious how your desire to
• 03:35do research was received by anymore.
• 03:39Dynamically minded clinicians.
• 03:43Well, you know. I mean, I had
• 03:45really been recruited to the program
• 03:46because of my research background.
• 03:48The reason my board scores?
• 03:52And it wasn't like I like the
• 03:54clinical, but you know, I think
• 03:56it was understood that you know.
• 04:00I was going to be doing some
• 04:02research at throughout it,
• 04:03but I think I tried to, you know nonetheless,
• 04:07like really immerse myself in
• 04:09the clinical programs and I. So.
• 04:14You know, I think people were supportive
• 04:18in general of the research. And.
• 04:23You know, I think you you try to be a
• 04:26good citizen and do all the clinical
• 04:28work and really kind of try to learn from it.
• 04:31While also not letting that other
• 04:33part of your brain totally turn off
• 04:36and trying to make sure that you
• 04:38take it seriously enough to wear,
• 04:40like if something really important,
• 04:42then you're going to prioritize it
• 04:43an even if it causes you know people
• 04:46to give you like some feedback like
• 04:48**** you know you kind of have to
• 04:51learn how to make those decisions
• 04:52for yourself and work with your
• 04:55allies and your program to get the
• 04:57support 'cause it's you don't want
• 04:59to be a situation where it's just
• 05:01you arguing against everybody else.
• 05:03But I was fortunate at the time to have
• 05:06been recruited by a program director
• 05:08who wanted to make research more of a
• 05:11conspicuous part of the training program.
• 05:13And so even though I faced some
• 05:15obstacles like I had my own initiative,
• 05:18I was able to create the projects.
• 05:20But then as I got through residency
• 05:23because I, like many right,
• 05:24like I had a really hard time
• 05:26getting much done during residency.
• 05:28Yeah, sure,
• 05:29you know the stuff that I was
• 05:31sort of collecting by just.
• 05:33You know, leveraging those lab resources,
• 05:36but.
• 05:39So after the residency was over,
• 05:41I helped them to compete for another 25,
• 05:45which at the time the program.
• 05:48Didn't know about that mechanism,
• 05:50so you know over the course of participating
• 05:54in some of those opportunities
• 05:56like the NIH is brain camp and.
• 05:59Some of their programs I I learned
• 06:01about that mechanism and then.
• 06:05I think for me it was appealing
• 06:07to take on a role like that so
• 06:10that in addition to the research.
• 06:13You know, coming out of residency,
• 06:15you knew you were going to have to
• 06:17piece things together with sort
• 06:19of additional responsibilities.
• 06:20But if for me I could take my
• 06:23experience having kind of navigated
• 06:24this complex landscape and sort of
• 06:27quantify that in a program that would
• 06:29be both like enjoyable for me and I
• 06:31could begin using that to both to
• 06:34find students and also just kind of
• 06:36continue learning as I was trying
• 06:38to now compete for my own award.
• 06:41And things like that so.
• 06:44So I helped him to get that and
• 06:46they were successful at it.
• 06:48And then I used to run the
• 06:51program for a few years and.
• 06:53You know, I think.
• 06:55One of those things where it's
• 06:57it's nice to see your programs
• 06:59continue to kind of take off on
• 07:02their own and then was able to kind
• 07:04of as I was getting other funding
• 07:06or other projects came along.
• 07:08I was able to sort of.
• 07:11Take a less directly involved
• 07:13role and then kind of gradually.
• 07:16I'm still involved with the program today,
• 07:19but.
• 07:21I guess something else that I've
• 07:23been skeptical of during my
• 07:25training and you mentioned this.
• 07:27Like how do we know you know, right?
• 07:30So I can terms of symptom assessment.
• 07:33A lot of your work now is
• 07:36measuring different symptoms,
• 07:37like getting back to like the
• 07:40kernel behavior or whatever and.
• 07:43At what point in your training did
• 07:45you start to wonder like whether
• 07:47even the words that we were using to
• 07:50describe conditions or like those
• 07:52sorts of dependencies that may or
• 07:54may not add up to bipolar disorder?
• 07:58What was your like journey through that?
• 08:02Yeah, I mean I think. You know,
• 08:06I guess my experience of that was sort of.
• 08:10It didn't really make much
• 08:11sense that that was how we were
• 08:13doing the evaluations, but.
• 08:16You know when you're training to
• 08:18become a doctor or psychiatrist,
• 08:20big part of that is just,
• 08:22you know what's the protocol?
• 08:23What do you need me to do?
• 08:25Should you need me to ask these questions?
• 08:28OK, you know, write down what they say.
• 08:31OK, like, OK,
• 08:32you're calling that pressured speech,
• 08:33OK, you're calling this low mood or
• 08:35you're calling this constricted affect.
• 08:40So I just saw it as a.
• 08:42We're just being trained to follow protocol.
• 08:45I'm not going to question,
• 08:47you know like on the one hand,
• 08:49like it's seemed really arbitrary and.
• 08:53Probably not biologically based,
• 08:54but at the same time it's.
• 08:57It's it's allows first kind of reliability
• 08:59that had a pragmatic utilities,
• 09:02so I really try to do separate in my mind,
• 09:06the pragmatic utility piece from.
• 09:10From but you know,
• 09:12but at the same time each.
• 09:14Experience of this of, like you know,
• 09:16why are we doing this again?
• 09:17OK, just that's fine,
• 09:18but just tell me you know.
• 09:20Through each clinical experience,
• 09:21just kind of noticing the places where
• 09:24something could be more objective.
• 09:28And kind of filing that away a little bit,
• 09:30you know, just to say like I can't possibly
• 09:33study everything right now, but like here,
• 09:35here's a way for me to say like.
• 09:39You know this one is really kind of fuzzy,
• 09:41and none of the clinicians know what this is,
• 09:44but they're constantly
• 09:45having to put it on paper.
• 09:48And it's causing a lot of confusion or
• 09:50like something like why are we documenting
• 09:52endzeit exactly like what does that mean?
• 09:55When I show a patient my note
• 09:56that says he has poor insight,
• 09:58he gets really upset.
• 10:00So like, should we be using a
• 10:02different word than insight?
• 10:03You know, I know what we mean,
• 10:05but like you know, just.
• 10:08Starts you thinking along the lines of.
• 10:10Like are these assessments
• 10:11were doing truly optimal?
• 10:12You know I didn't.
• 10:14It's like and you kind of
• 10:15know that they're not.
• 10:17But you know that they're sort of time
• 10:19honored and there's not a lot of evidence.
• 10:22Do something different,
• 10:23and so you know it's question is like.
• 10:27If we are going to change things,
• 10:28how would we know that we're changing
• 10:31them for the better and stuff so?
• 10:33But
• 10:34the question I think
• 10:35it's interesting 'cause not
• 10:36everyone thinks along those lines.
• 10:38Alright, so I'm wondering if.
• 10:40I mean, I I I've noticed many residents
• 10:43don't don't think along those lines.
• 10:46Many attendings you know people have
• 10:49been practicing for their entire career.
• 10:51Don't really question that. So.
• 10:53So there's a difference between becoming
• 10:56proficient at detecting pressured speech,
• 10:58say and wondering what exactly
• 11:00is pressured speech like.
• 11:01At what frequency of words does it
• 11:04become pressured from normal or rapid?
• 11:07Or like where is the line? Yeah,
• 11:10yeah, and I wasn't.
• 11:11I mean, I wasn't necessarily
• 11:14preoccupied with that like in terms of.
• 11:18Defining the words, but I think.
• 11:22I guess it was more about
• 11:25when you go from the.
• 11:27Stage of training where you're
• 11:28really just filling out the forms
• 11:30the way you know to fill them out
• 11:32can work done to get the work done
• 11:35to really trying to get to be better
• 11:37at it to be more efficient at it.
• 11:39To kind of have a more intuition
• 11:41around you know these kind of master
• 11:43clinicians who could come into a room
• 11:45and then within a few seconds have
• 11:47zeroed in on some core pathology to
• 11:49me that was really fascinating, right?
• 11:51Like you know we spend so much
• 11:53time getting these notes documented
• 11:55for billing and all these things.
• 11:57But you know you have the clinicians
• 11:59who are not doing that,
• 12:01but they're able to kind of come in
• 12:03and ask these incisive questions
• 12:05and get
• 12:06to the heart of the matter.
• 12:08Something I remember observing
• 12:09in intern year was there's a.
• 12:11There's a clinician, Tom Duffy.
• 12:13Here, who is a hematologist.
• 12:15I've ever one morning was like
• 12:177 in the morning we've been
• 12:18this is a medicine rotation,
• 12:20even fretting over this one patient for it.
• 12:23Half an hour you know the whole team
• 12:27standing around and he came in and.
• 12:29Within 30 seconds it couldn't
• 12:31have been more than 30 seconds.
• 12:33He knew what was wrong.
• 12:35He ordered the tests and
• 12:36the test came back exactly.
• 12:38See predicted and I remember thinking like.
• 12:41So here's a guy who was detecting some
• 12:45signal which none of us were able to detect.
• 12:48But then there was another step
• 12:50where he was able to demonstrate
• 12:53that what he had detected is
• 12:55accurate and his prediction.
• 12:57He had made a quantifiable prediction
• 12:59with Mary Unquantified an I've had
• 13:02the same experience in psychiatry,
• 13:04where people with equal vigor can
• 13:06conviction state a formulation for case.
• 13:08But then there's no way to.
• 13:11Really test whether that's accurate.
• 13:16Well, right? I mean sometimes it is.
• 13:19You know you can ask them,
• 13:20and that's like if you have a
• 13:22sense that like I bet this is
• 13:23somebody with trauma history.
• 13:24I'm just getting that vibe or there's
• 13:26something I'm picking up, sure. Yeah yeah.
• 13:27Then you can ask for you start to use your.
• 13:30Hypothesis generation to basically
• 13:32gradually zero in on that, so I think.
• 13:37So I think we we still, as you know,
• 13:40good clinicians still do that which is
• 13:43like within a very short amount of time.
• 13:46They use the gestalts, sort of.
• 13:49Where you're looking,
• 13:50how you're moving to generate some
• 13:52hypothesis and then the questions
• 13:54are really kind of designed to zero
• 13:57in on that mythology and then.
• 14:02You know? Is there like a lab
• 14:05test you can then run to be 100%?
• 14:08Generally no, but.
• 14:11Our tactic is usually like,
• 14:12well, if I'm right,
• 14:13then I should be able to use this
• 14:16medication and then it will get better.
• 14:18So yeah, it's it's a gap that
• 14:22needs to be filled. Well.
• 14:24Occurs to me now that a lot of your
• 14:27research in the digital phenotyping is more
• 14:29precisely defining the problem. Set right?
• 14:32So like you work with accelerometers,
• 14:34speech analysis, facial expression like
• 14:36these are all things that you can do.
• 14:39Like yeah, I mean,
• 14:40well obviously not the accelerator.
• 14:42Maybe your Geo location,
• 14:43but certainly you can look at a patient.
• 14:47Analyze their face,
• 14:48see where their eyes are gazing.
• 14:50Don't tell something about their
• 14:52body language or affect and you
• 14:54don't need a number for that, but.
• 14:57That maybe maybe you're trying to pin
• 14:59yourself down to a number like cannot?
• 15:01Could I understand it like that or is?
• 15:04Yeah,
• 15:04I guess the way I would think about
• 15:06it is like my experience of being
• 15:09a psychiatry resident was like.
• 15:11You're not very good at everything.
• 15:13You're being trained to do 'cause
• 15:14you're just learning, and so you get.
• 15:16But you get exposed to people
• 15:19who are really good at it.
• 15:21And then you're trying to see if you can
• 15:24figure out how to get good like that,
• 15:26and what exactly is it that they're doing.
• 15:29That's different than what I can do, and so.
• 15:34The ability of really good
• 15:36psychiatrists to enter ologist to
• 15:38pick up on these subtle things.
• 15:41And like be able to tell that this
• 15:43particular type of movement is a
• 15:45lithium trimmer because it's in this
• 15:47frequency or that this particular
• 15:49kind of head nod is Parkinson's
• 15:51versus essential tremors because of
• 15:53sort of like the precise dynamics
• 15:55or the way that it's moving.
• 15:59Or that this person speech is
• 16:02manic and this other persons is
• 16:04psychotic because of very subtle
• 16:07dysarthria is that they're hearing?
• 16:10It was that being able to map between these
• 16:12sort of subtle constellation of features
• 16:14into sort of a much more coherent formula.
• 16:17That to me was what was really
• 16:19cool about it, which is like.
• 16:21I can kind of see that they're doing this,
• 16:25but like I am not good at it and this idea
• 16:28that I'm just going to like see a million
• 16:31patients and eventually get good at it.
• 16:34Seems to me kind of crazy, because then.
• 16:38Like no one will get good until
• 16:40they've seen people for 10 years.
• 16:41Yeah, what about the half
• 16:43a million they see before
• 16:44their angry and
• 16:45so and so? How is it that you know
• 16:47going to school in a teaching hospital?
• 16:49You know you're not providing great care,
• 16:51but you're kind of like wow,
• 16:53they're letting me see people even though
• 16:55my skill set is so mature at this point.
• 16:59And just kind of feeling uncomfortable about
• 17:01that and thinking like, gosh, you know,
• 17:03if there was at least some ways of.
• 17:06Having some assistance in terms of some of
• 17:08these features which should be quantifiable.
• 17:11Like, shouldn't we be investing
• 17:12in that kind of thing?
• 17:14Even so that like somebody like me
• 17:16could learn to do it way faster or
• 17:18there wouldn't be as much liability
• 17:20when I'm not trained up and stuff
• 17:22or something as basic as like I'm
• 17:24going to have to go to my supervisor
• 17:27later and tell them how it went?
• 17:30And if I feel like dodging it,
• 17:32I could just talk about neuroscience.
• 17:35Alright baby.
• 17:37Or if I really want to learn that day,
• 17:40I could record the session and then
• 17:41play it from my supervisor and we
• 17:43could go through it in great detail
• 17:45and I'll feel really embarrassed.
• 17:46And yet, like I will learn way more that day,
• 17:49could they let you
• 17:50do that? Sure? Oh,
• 17:51they did not let us record.
• 17:53I would have loved to have done
• 17:54that. I mean, I think in most programs
• 17:56it's considered standard of care.
• 17:58If you're learning psychotherapy
• 17:59to record your sessions, not a lot.
• 18:02Yes, that's unusual. I think.
• 18:04I mean, it's a it's definitely part of
• 18:07almost every psychological training.
• 18:08And then yeah, no. We were.
• 18:10We were strongly encouraged.
• 18:12You know, with patient permission,
• 18:13you gotta get them signed.
• 18:15Yes, to get audio recordings
• 18:17and in some cases like video
• 18:19recordings of your sessions,
• 18:20because unlike the old days where
• 18:22you would have a one way mirror
• 18:25an you get somebody like really
• 18:28like watching you and Nikki notes.
• 18:30So here was a way to use technology
• 18:32in a very simple way,
• 18:34which is just don't worry bout
• 18:36scribbling down your process notes.
• 18:37I mean there may be a reason to do that,
• 18:40but let's let's have you actually
• 18:42record verbatim what was said and
• 18:44your posture and all the things?
• 18:45And and let's just look at it
• 18:47and see what you might be doing
• 18:49differently and stuff.
• 18:50And to me like those were
• 18:52the moments that were both.
• 18:53Again like you feel yourself being
• 18:55sculpted out of out of stone.
• 18:57'cause it's like kind of painful.
• 18:59But at the same time you're like.
• 19:01Oh,
• 19:01that's what you mean by this and you
• 19:04kind of have somebody's not in the room,
• 19:06but it kind of in the room to train you.
• 19:09So it was all sorts of variances
• 19:12that to me it was like this.
• 19:14OK,
• 19:14obviously it would be way more
• 19:16efficient if everyone had to
• 19:18record every single session, right?
• 19:19Because I you know,
• 19:20like and I get that there there are
• 19:23technical and privacy issues with that.
• 19:25And like not every patient may want to do it,
• 19:28but. It just seemed like a natural thing.
• 19:31If you're going to be in a training
• 19:33hospital that for the types of
• 19:35encounters where they really
• 19:37couldn't be someone in the room
• 19:39for various reasons that you needed
• 19:41systems to be able to objectify,
• 19:42like what was going on.
• 19:44If only the if the only reason to do
• 19:46it was to get more useful supervision
• 19:49and not be able to like Dodge,
• 19:51sure your blind spots
• 19:52stuff. I'm wondering so it sounds like
• 19:54there's a different orientation towards
• 19:56technology then, at least in the.
• 19:58The long term care.
• 20:00Training some aspect of the residency
• 20:03training between our programs and I've
• 20:07wondered sometimes if the culture
• 20:09of an institution is such that.
• 20:11People don't want measurement
• 20:14because then they could be disproven.
• 20:17And so I wonder,
• 20:18so some some attendings that one of the
• 20:21reasons I really respected Doctor Duffy.
• 20:23It was he would tell you what his
• 20:25prediction was and that way you
• 20:27know it and he knew it and he
• 20:30was testing himself and holding
• 20:32himself accountable to prediction.
• 20:33I haven't found it.
• 20:35That's the case as much in psychiatry
• 20:37and I've been curious whether that's
• 20:39a cultural thing that's pervasive
• 20:41or maybe just my way of eliciting
• 20:43bad reaction from people, yeah?
• 20:46Yeah, I mean, I think the culture of
• 20:48measurement and like psychiatry's
• 20:50is an interesting one I mean.
• 20:54Even something as basic as.
• 20:57When you have a complex
• 20:59patient being willing to get a
• 21:01neuro psych evaluation where,
• 21:02like the neuropsychologist could
• 21:04come in with their battery of tests
• 21:06and provide you like a system by
• 21:09system breakdown of their capacities.
• 21:13Which. Psychiatrist Azharul didn't
• 21:15didn't do, and part of that was.
• 21:19Well, is it really going to change
• 21:22my management or you know it's
• 21:24a lot of additional valuation?
• 21:25I'm not sure it's
• 21:27well so that so that question
• 21:29there is interesting, right?
• 21:30Because will it change my management and?
• 21:33I have had the experience that
• 21:36there isn't much that would change
• 21:38some people's management right now,
• 21:39and no amount of data
• 21:41will move amount right, so
• 21:43I think I think rather than
• 21:45being prescriptive of like woman,
• 21:47of course you should measure
• 21:49because this medicine and come on
• 21:51guys like let's measure the brain.
• 21:53Let's measure behavior.
• 21:54You know it's a no brainer,
• 21:56so to speak. But I think.
• 22:00One of the other really important
• 22:03skillsets I think of becoming a
• 22:06psychiatrist is sort of working with.
• 22:08Resistance when someone doesn't
• 22:10want to change their behavior?
• 22:11Sure, yeah.
• 22:12And then understanding that in terms
• 22:14of like not just being like will
• 22:16come on like why won't you change?
• 22:18You know, but to say like that's interesting.
• 22:21So what exactly is it about
• 22:23this that you would think is
• 22:25not worth doing this really to?
• 22:29To roll with the resistance right and
• 22:31to try to understand what is it about
• 22:34that additional thing that if you
• 22:37ordered that tasks and you got it back.
• 22:39You wouldn't want to necessarily use
• 22:42the data, so I think I learned a
• 22:44lot around just trying to, you know.
• 22:48Just accept that resistance as a valid
• 22:50thing and they do think it exists and
• 22:53it's valid and we can't dismiss it.
• 22:55And to think back around well, what?
• 22:58Why is that?
• 23:00You know, is it still worth measuring?
• 23:02How do we address the the places
• 23:05where there is a particular concern,
• 23:07whether it's a privacy concern or
• 23:09I don't want to be have my judgment
• 23:12usurped by this test kind of concern.
• 23:17'cause then it's sculps.
• 23:19Sort of how you think about incorporating
• 23:22the measurement and like where in
• 23:25the clinical decision making process
• 23:27there maybe is a role for a tasks
• 23:30that help somebody reduced there.
• 23:34There you know.
• 23:36That the uncertainty in situations where.
• 23:41I really don't know what to do
• 23:42and there might be a place that
• 23:44something like that would be helpful,
• 23:45but. Soon
• 23:47very effective at doing that,
• 23:50approaching people.
• 23:53Psychiatrist of understand.
• 23:54Recognizing the resistance and
• 23:56then being able to navigate it.
• 24:00You even successful at implementing
• 24:03these digital phenotyping
• 24:04procedures on different units,
• 24:07which seems. Like a real
• 24:10cool well I don't know how
• 24:12successful you necessarily bent.
• 24:14I would say that my experience of of
• 24:16trying to get these kinds of measures
• 24:19into clinical services has been
• 24:21very different from my experience.
• 24:23Is trying to get a neuroscience perspective
• 24:26into those same clinical services.
• 24:29Well, in other words, like you know,
• 24:31going to the psychosis unit and
• 24:33saying like you know we should be
• 24:35scanning everyone who comes here
• 24:37with a functional scan because I
• 24:39bet we can find that there's this
• 24:41difference in their brain that we
• 24:43should then use and part of our
• 24:45evaluation and people are like, OK,
• 24:47we like the idea of the brain thing,
• 24:49but like what exactly is it that you
• 24:51need to do and how is that going to
• 24:54change what I'm doing for this person?
• 24:59And having to kind of really be.
• 25:02Humbler, Alec would have,
• 25:03like you know, that's a good point.
• 25:06I'm not sure it would change anything in my,
• 25:09you know, I would love the data
• 25:11you know from a researcher,
• 25:13but you know, you're right.
• 25:14Like I guess I can't tell you exactly
• 25:17how you would use it in a way that you
• 25:20know you might not be able to kind of get
• 25:23that information from their behavior.
• 25:25So the idea of sort of getting psychiatrists,
• 25:28whether it's residents or attendings
• 25:29wherever to like really care about the
• 25:32underlying biology of what they're seeing.
• 25:34Um, you know, I taught some of the clinical
• 25:38neuroscience curriculum for many years and.
• 25:40You know it's really variable.
• 25:42Some some people are really
• 25:43interested in it somewhere,
• 25:45just like totally glaze over.
• 25:48And initially, you're you perceive
• 25:49that as sort of threatening,
• 25:50like why can't these people care
• 25:53about their organ of interest?
• 25:55It's so irresponsible,
• 25:56you know as things,
• 25:57but then you realize you know as
• 25:59you go through the training program.
• 26:01Like psychiatry is really hard to do well.
• 26:04Just with the tool you know the tools
• 26:06that we have and if I'm providing a tool
• 26:09that provides no additional information
• 26:11or or help that person's day go smoother,
• 26:14or you know,
• 26:15or you're asking him to learn a
• 26:17whole new field of information.
• 26:19Like of course they should be sceptical,
• 26:21right so?
• 26:22The other experience I had done that for many
• 26:25years and tried to get you know neuroscience.
• 26:29You know into the minds of the
• 26:31did you make playdough brands?
• 26:34But like you know I,
• 26:36I did try to help teach.
• 26:38You know what I considered?
• 26:40Neuroscience 101 that every
• 26:43psychiatrist should sort of know.
• 26:45And I think I still do that to some extent,
• 26:48although my approach has changed quite a bit,
• 26:50but.
• 26:52When we started getting more into the.
• 26:56The digital phenotyping where we were
• 26:59taking people's behaviors and really
• 27:03trying to study them more precisely.
• 27:06The approach was quite different,
• 27:08which was more of that.
• 27:11You could go to the clinicians and say
• 27:13look you're an expert at reading the
• 27:15behavior I need your help to design a
• 27:17system that can do as well as you can,
• 27:19or even just pick up on some
• 27:21of what you're picking up on.
• 27:24And like I had already trained with
• 27:26many of these people, so I like,
• 27:29I knew, OK,
• 27:30this person's got an amazing ability
• 27:32to pick up on those subtle trimmers,
• 27:34right?
• 27:35Or those subtle dysarthria's,
• 27:36or those little movements of the face
• 27:38that in the context of an interview they
• 27:41could infer was a sign of paranoia right?
• 27:44And so to me that was fascinating
• 27:47that they had this ability,
• 27:49but I wasn't sure how much of it was
• 27:52real and how much of it was superstition.
• 27:55But what was great about it was I
• 27:57could go to them and not say, like,
• 27:59hey, I've got this technology.
• 28:00I'd like you to start using it.
• 28:02I went to them to say, hey.
• 28:03I really need to find ways of measuring this.
• 28:06Can you help me design a system?
• 28:10Mission as you are? Yeah yeah.
• 28:12And so the clinicians loved it because
• 28:15they were like this is really cool.
• 28:17I've always wondered if
• 28:19what I'm hearing is real.
• 28:22And we could begin connecting
• 28:24them with computer scientists who
• 28:26were expert at decoding audio,
• 28:29audio, or speech signals
• 28:30from audio and and decoding.
• 28:33You know,
• 28:34facial movements from video and
• 28:36linking those people who were really
• 28:39interested in the mental health
• 28:41aspects but had a hard time gaining
• 28:44access to the data with the clinicians
• 28:47who were really fascinated by the
• 28:50nuts and bolts of behavior which.
• 28:53Which was not everyone but the certain
• 28:54clinicians had that inclination.
• 28:56And and then just getting those
• 28:57two groups to be able to talk to
• 29:00one another basically and being
• 29:01the translation element of saying,
• 29:03you know, I consider myself like a
• 29:05mediocre psychiatrist, but I can,
• 29:07at least I know what she's pulling out.
• 29:09And I know that.
• 29:10Here's how we were taught to think about it,
• 29:13and so let's design some systems
• 29:15that can pick up on it.
• 29:18Hydrus loved it.
• 29:19The patients loved it.
• 29:20Unlike, you know a brain imaging experiment.
• 29:23You know,
• 29:24we pay them.
• 29:25But like we're making them sit in this really
• 29:28loud tube and play these boring video games.
• 29:31Whereas like the experiments to study
• 29:34patients in these interactions with the
• 29:36doctor was for them really easy and fun.
• 29:39You got to talk to somebody and
• 29:42talked about your problems.
• 29:44We're lining up to do the studies
• 29:46and so just created are different.
• 29:49Kind of you know,
• 29:50right there on the ground on the unit.
• 29:52We didn't have to leave the unit,
• 29:55we could just bring people into
• 29:57these rooms where it felt like,
• 29:59hey, this actually is like.
• 30:01A sustainable way of digging
• 30:02into the pathology which no one
• 30:04else seems to be really doing.
• 30:05So taking that thing where you just
• 30:07sort of go to a clinician and say,
• 30:10alright, you've got a lot of pearls.
• 30:12You know you taught a bunch of
• 30:13them to me during residency,
• 30:15but let's see if we can like build a
• 30:17computer that operationalize pearls,
• 30:19and then in my in the back of my mind.
• 30:22I'm thinking like what we really
• 30:23want to do is design something
• 30:25that if we could scale that up,
• 30:27we would be able to know,
• 30:29like which of these things are true pearls.
• 30:31Yeah, and which of them are just,
• 30:34you know,
• 30:34like rocks or what's right like that
• 30:37kind of they look good but they're not,
• 30:39but they're not actually correlated with
• 30:41the outcomes that we think they are so.
• 30:46And I think that including the clinicians
• 30:48in that process to say, like, hey,
• 30:50look, you know we're all fallible.
• 30:52This is, you know,
• 30:53this is what you were taught,
• 30:54so we want to get back to sort of.
• 30:58You know you've honed your interview
• 31:01to be efficient, but let's see if
• 31:03we can make it even more efficient.
• 31:06Let's see if we can help you
• 31:09to train residents or the next
• 31:11generation with these videos as well.
• 31:14As you know,
• 31:15your kind of classroom didactic style.
• 31:19Well, it seems like it's a very
• 31:21effective way of doing it.
• 31:22I think it could be it.
• 31:24I think it's still evolving and I
• 31:25think it would be great if there
• 31:27were a whole platform around.
• 31:29Sort of when you come into
• 31:30a training environment.
• 31:31This is just the expectation,
• 31:33which is that like you're going to be
• 31:35learning from a lot of pre existing video,
• 31:37you're going to be learning from
• 31:38a lot of encounters from your
• 31:40mentors that are being recorded.
• 31:41You're going to learn from your
• 31:43own encounters that are being
• 31:45recorded and then whether or not
• 31:47the mentors in the room with you.
• 31:49There will be all sorts of
• 31:51statistics about your interview
• 31:52that they can look at and be like.
• 31:54Oh yeah,
• 31:54now you really don't want to smile
• 31:56the whole time like that's going to
• 31:58or not small.
• 32:02And it's been fascinating in the labs.
• 32:05Now we record the dyads where the
• 32:08research assistants and stuff are.
• 32:10Interviewing the patients so we can
• 32:12see what somebody looks like when
• 32:14they're fresh off the training,
• 32:16you know. Just start talking
• 32:19to patients like fresh off the
• 32:21medicine wards or just coming. You
• 32:23know, new research assistant joining the lab.
• 32:26No clinical experience and like how do they?
• 32:31Conduct themselves in encounter
• 32:32where the you know what are the
• 32:35kind of intrinsic skills that some
• 32:37people bring to that that there is,
• 32:39you know they can build report quickly,
• 32:42let's say. Or where you know
• 32:45where does that get in the way?
• 32:48Like if you're trying to create too
• 32:50much for poor and therefore like
• 32:52people don't ever quite exposed
• 32:54their pathology in some ways.
• 32:55So it's yeah, it's just interesting to.
• 32:58In that context I can require all the
• 33:01research assistants to do it this way,
• 33:03and then we can see as someone
• 33:05who gets better at it,
• 33:06like what are they doing differently?
• 33:08And I
• 33:09just had this idea that you have a
• 33:11lab or and you are experimenting on
• 33:14how best to train a psychiatrist.
• 33:16For your research assistance,
• 33:18so they're like your.
• 33:20I don't know just Guinea pigs,
• 33:22but like her subjects and
• 33:24you're changing the experimental
• 33:25conditions and then are you, are you
• 33:27still involved in the residency program?
• 33:29So you're translating that straight into the.
• 33:32Training of like MD physician? Yeah
• 33:34no. I mean that that would be cool.
• 33:36I mean I guess my.
• 33:39You know, if they were to ever
• 33:41come to me to say like hey Justin,
• 33:43here you have this great training
• 33:44thing, can we use it?
• 33:45I would be like sure, let's try it.
• 33:48When we've tried to explore
• 33:50even using the pre recorded
• 33:51videos for educational purposes,
• 33:53the IRB is basically said like Nope
• 33:56and I think that has to do with like
• 33:58a lot of the ethics of sort of when
• 34:01you're engaging in research studies.
• 34:04Who's going to use that data in one of
• 34:07the context that are considered appropriate?
• 34:10I think down the line that may happen, but.
• 34:13You know, I'm not trying to
• 34:15foist it on them at this stage.
• 34:16There's a lot of other legitimate reasons
• 34:18why they may not want to go that route.
• 34:21But just back to your kind of other
• 34:24earlier comment like you know now it
• 34:27has sort of evolved into thinking about
• 34:30the lab as a little sandbox of you know,
• 34:34different health care,
• 34:35delivery and training.
• 34:39You know systems where you know what if
• 34:42it were possible to train somebody up to
• 34:45be really competent at doing interviews,
• 34:48both for information extraction
• 34:49like just doing some valuation,
• 34:51but also potentially providing therapy,
• 34:53like if there were ways to short circuit
• 34:56this apprentice based system by having a
• 34:59much more tech enabled feedback system
• 35:01where your supervisor didn't have to watch,
• 35:04like every hour of every video but like.
• 35:08It could identify features that you
• 35:11know start to gain some some trust
• 35:16around those features that helps a.
• 35:19More experienced clinician.
• 35:21Really read the report,
• 35:23kind of what you did and then
• 35:25give you more rapid feedback.
• 35:27And then you're trying again
• 35:29and you can gradually.
• 35:31And maybe even rapidly improved.
• 35:34So yeah, I would have loved that.
• 35:38Is really cool.
• 35:39What is this one of these like?
• 35:42You kind of are amazed at how
• 35:44inefficient you know the training
• 35:46process is right and it's like as
• 35:49somebody who is concerned about the
• 35:51overall cost of health care and
• 35:53mental health in particular and
• 35:55and all of those kind of high level
• 35:59policy things like it seemed to me.
• 36:01Pretty egregious how inefficient
• 36:05our training was considering.
• 36:09That that basically bakes in a lot of costs,
• 36:12right? And that like it's not,
• 36:14it's not really up to us whether that's.
• 36:18That's just the way we do it,
• 36:20you know.
• 36:21I think you know you kind of have
• 36:23to sort of see the way things are
• 36:25moving in terms of value based care,
• 36:27measurement based care.
• 36:28If like you're not going to get
• 36:30to spend every hour of,
• 36:31we can let you know doing evaluations
• 36:33like it has to get more efficient.
• 36:34You have to be able to demonstrate.
• 36:37Why each question you ask is actually?
• 36:40A good use of your time,
• 36:42because if it were a lab test or something,
• 36:45right?
• 36:45Because if you don't,
• 36:47if you're not able to at least
• 36:49demonstrate that value other people
• 36:50are going to come in and do you
• 36:53know more mediocre valuations,
• 36:54but the outcomes will be fuzzy
• 36:56and insurance companies or payers?
• 36:58We're going to ultimately say well,
• 37:00we're only going to pay for that
• 37:02thing because this much more detailed,
• 37:04nuanced valuation costs way more and
• 37:06doesn't seem to be that more effective,
• 37:09so.
• 37:10I think the same for training
• 37:13like if we can't figure out how to
• 37:16train people more efficiently than.
• 37:18It could end up actually being
• 37:22problematic for the.
• 37:24Yeah, for the field you know so.
• 37:28Well,
• 37:28I wish you all the best doing that and
• 37:31maybe I'll go through residency again
• 37:33if you roll out your tech based open
• 37:36course on how to be a psychiatrist.
• 37:38Or you can
• 37:39help you know. Disseminate it.
• 37:43Once it's out there.
• 37:44Very happy. Yeah, I'd be
• 37:45very happy to thank you. Thank
• 37:47you so much for taking the time to
• 37:49talk, that's great, thanks.
• 38:01Well, I hope you enjoyed that episode.
• 38:04Thanks again to Justin for being
• 38:06on the podcast and you can find
• 38:08Justin on Twitter at Justin Baker,
• 38:10MD's in medical Doctor.
• 38:11Again, that's at Justin Baker MD.
• 38:13You can also find him on his
• 38:16partners.org faculty profile
• 38:17page or on Google Scholar.
• 38:18Just look up Justin Baker thanks
• 38:20to the Yale School of Medicine for
• 38:23sponsoring the podcast and especially
• 38:24to Adrian Bonding Burger for.
• 38:26Producing the podcast and Ryan
• 38:28McEvoy for his help sound editing.
• 38:29A special thanks to you for
• 38:31listening and again,
• 38:32my name is Daniel Barron and I've
• 38:34been your host and I'll see you
• 38:36next time here on science at all.