January 24, 2025 • 50 mins
Can artificial intelligence revolutionize the way therapists document and manage their sessions? Join us as we unravel this compelling question with Dr. Melissa Mcafree, a leading voice in mental health innovation. You'll gain a deeper understanding of how AI is reshaping clinical documentation, balancing detail with efficiency, and addressing concerns like HIPAA compliance. Dr. Mcafree's insights will illuminate the path to harnessing AI as a powerful ally in maintaining accurate and secure progress notes, ultimately enhancing the therapeutic process.
We venture into the realm of AI-assisted progress notes from telehealth sessions, exploring both the potential and pitfalls these technologies present. Learn how thoughtful integration of AI can alleviate the administrative burden on therapists, while ensuring that session notes remain true to the client's experience. Through candid discussions, Dr. Mcafree emphasizes the importance of informed consent and ethical considerations when weaving AI into therapeutic practice, ensuring that client data remains protected and private.
As we discuss strategies to enhance documentation practices, discover how a hybrid approach of personal input and AI assistance can lead to stress-free, efficient note-taking. Dr. Mcafree offers actionable tips and resources, including her book "Stress-Free Documentation for Mental Health Therapists," equipping professionals with the tools they need to succeed. Whether you're a seasoned therapist or just starting out, this episode promises to expand your understanding of AI's role in therapy and inspire confidence in embracing these cutting-edge solutions.
Get your step by step guide to private practice. Because you are too important to lose to not knowing the rules, going broke, burning out, and giving up. #counselorsdontquit.
Mark as Played
Transcript
Episode Transcript
Available transcripts are automatically generated. Complete accuracy is not guaranteed.
Speaker 2 (00:06):
word and uh.
Welcome dr melissa mcafree andwelcome to uh texas counselors.
Creating badass businesses freeworkshop awesome, thank you.
Speaker 3 (00:18):
Thank you for having me, and you will have to um
excuse the unprofessionalpresentation today, meaning that
my cat went to the vet thismorning and was medicated and is
just kind of loopy and whining.
So she may make an appearanceand you may hear her, and I
apologize.
Speaker 2 (00:38):
That is totally okay.
Speaker 3 (00:40):
Thanks, hopefully you can't hear her in the
background right now.
No, I don't hear her.
Kat, you're good?
Okay, good, good.
But yeah, I saw that we had afew questions coming in and yeah
, so I'll let you take it fromthere and where we should go
next.
Speaker 2 (00:59):
Excellent.
Well, I just made you co-host.
So if you need to share screenor anything, let us know
Everyone else.
You know-host.
So if you need to share screenor anything, let us know
Everyone else.
You know the drill.
If you need CEs for thispresentation, leave your camera
on and we will post a link.
Oh, we just did Okay.
So there's the link.
It looks like I'm posting it,but it's actually my assistant,
jennifer, who's logged in as me,because you know wifi.
So here we go.
Speaker 3 (01:21):
Oh, awesome, and I I don't have a presentation, so I
thought this was more just a Q&A.
Oh no, no, absolutely.
So if we wanted to start outwith those, then that would be
great.
Or if you have anything inparticular you want me to kind
of riff on, I can talk forever,so please interrupt me.
(01:41):
I love this talk forever.
Speaker 2 (01:42):
So please interrupt me.
No, no, this is great, so Iwill.
I will ask you guys take 30seconds or less and put it to
the chat.
Why are you here?
Why are you here?
What are you hoping to learnabout documentation?
What is worrying you?
What is?
Does AI enter your brain?
(02:04):
Are you looking to AI assomething that could save the
day, or are you looking at it asTerminator 5 or whatever we
think AI might be?
So start posting.
Speaker 3 (02:16):
Let me see.
And yeah, and I could say thatI would say there are some few
common things that people reachout to me.
For Obviously, we can get asnitty gritty and in detail as
you want and we do sometimes butthe most common things that I
hear from people is like how doI balance?
How much is too much in aprogress note?
(02:36):
When am I writing too much,when am I writing too little?
Related to that, one thatpeople don't often publicly say
but is an extremely commonexperience, is how do I catch up
on my progress notes, because Ifell really behind.
Speaker 2 (02:52):
I don't know anything about that.
Speaker 3 (02:53):
Yeah, and I can't tell you how many people have
told me.
You know, please, you know Ihave to be the worst one that
you've seen.
You know, like there's no wayyou've heard how you know that
someone else is as behind as Iam and I'm like Nope, I have, I
promise you.
So I've met people who havenever done notes so hard to beat
(03:13):
that right.
So you know, it's such a commonconcern, though, and it's a
very shameful experience to haveand to admit, on top of the
whole experience of just notfeeling confident in
documentation and both of thosein my like anecdotal, my own
experience with working withpeople and I've been doing this
(03:34):
for 10 years now has nothing todo with the school you went to,
with the license type you have,with the amount of experience
you have.
I work with people who have beendoing this for decades and
people who are just fresh out ofgrad school.
So those are two really commonthings, and I would say that,
(03:54):
because you say we want to getinto the AI stuff, ai is a
problem or is a solution thatpeople are hoping for that
people think will solve eitherof those problems, right?
So AI will write my note betterthan I can.
Ai will choose interventionsthat I can't think up in my own
(04:15):
head.
Ai will write it in a clinicalway.
Ai will include the informationthat insurance wants to see, or
, you know, ai will help me dothis faster, and so I think of
those two really common issuesthat I work with people on
specifically for progress notes.
I think that's what theyusually want AI to do, and I
(04:38):
will be the bearer of both badand good news in that.
I have worked with people whohave said AI is like they've
used the phrase life changing,so it really can help, and you
still have to know what goes ina progress note to be able to
use AI, because let's startthere then.
Speaker 2 (05:01):
Yeah, I mean, like the first, I'm looking.
We kind of released.
You know we have tons ofquestions now, but I see the
first one.
I see HIPAA security and I knowyou'll talk about that, but
wanting to hear how AI canassist with clinical
documentation and you canaddress both of those things
right, the quality of the noteand can it help me catch up?
(05:22):
Right, those two things yousaid are kind of the most, and
this would be a good time tomention too.
You help people with this andyou just wrote a book, so you
have lots of resources for usand you will be put up there.
It is All right good, and youwill be putting the link in the
chat for us before we end todayas well.
So, yeah, talk about AI and howit can help with those two
(05:44):
things, or if it can, or what weneed to do to make it do that.
Speaker 3 (05:47):
Yeah, absolutely yeah , and I think we can combine
quite a few of these questionsinto a kind of long winded
answer.
But please continue puttingyour questions in the chat.
It is helpful if you putquestion and then a colon before
you write your question out, asfar as being able to see like
who's just following up onsomething, who's writing a
comment and who is actuallyasking a new question.
(06:08):
So for AI, let me back up.
So, yeah, we'll kind of talkabout how it can be useful, what
to look out for, and when wetalk about what to look out for,
we'll talk about HIPAA.
So, because it's actually a lotmore simple than you'd think,
that part of it.
So with AI, let me talk abouthow some of these work.
(06:33):
I think that's a good place tostart.
So there are various options.
You can log in all the way from.
The AI platform will offer likea telehealth platform and you
will do your session overtelehealth that way and it is
literally recording the entiresession, listening into the
whole session on both sides, andthen after you end your session
(06:57):
, it creates typically atranscript, because it's
listened to the whole thing, anda progress note for that
session.
That can sound both veryawesome and very scary for many
therapists.
So you know and both feelingsare relevant this I have so many
(07:19):
things because I do like awhole AI training on this.
So with that, I would say, mybiggest takeaway for anything AI
.
So, regardless of what I'mgoing to say in any of this
other stuff, and if you takeaway anything, you have to edit
the AI note.
You must.
You have to know what should bein that note.
(07:42):
So I'll give you some examples.
I use the same therapy sessionto test a lot of these AI
platforms and I have a freeYouTube series where I'm testing
all the platforms.
I had to take a break while Iwas finishing up my book, so I'm
just now getting back intotesting more of them and needing
to retest them because theychange so frequently that now, a
(08:05):
year later, some of my reviewsare almost obsolete because they
have new features.
Every single one of them hasmade up information, every
single one has gotten somethingwrong and I would say my biggest
critique of overall, all the AIplatforms is that they are
extremely wordy.
(08:26):
They give you these huge giantsummaries of information and so
time-wise for most people.
Ai doesn't really save timebecause you have to the time to
read through the mountain oftext that it gives you and then
either delete it or even justsay like it's perfect as it is,
(08:47):
which I have never seen, I willtell you.
But let's say you only had tochange one thing.
It really does take a lot oftime.
Now that doesn't mean it's notworth trying out if you really
struggle with notes, because ifit saves you the mental effort
sometimes, if it takes the sameamount of time but you are not
needing to sit there and producesorry, if you can hear my cat,
(09:10):
she's being very loud If you'renot needing to produce all of
that information from your head,that can be a huge lifesaver.
I mean, that can be dramaticand it might mean the difference
between you avoiding writing anote and you feeling comfortable
logging in and getting it doneif it doesn't feel like it'll be
a lot of work.
So I wouldn't say that's a.
(09:33):
It's not necessarily a drawback, it just is a fact.
The other thing is that most ofthese AI platforms are much
slower.
If you have used something likeChatGPT for other things,
that's not how this is going towork.
You're going to hit submit oryou're going to end your session
and you're going to literallylike they'll even say, go grab a
coffee, enjoy doing somethingelse while we write your notes.
(09:56):
So I mean, it takes a while forthese systems to produce a
progress note for you.
So you know, especiallyespecially if they're processing
like an hours long session.
So it's not like end yoursession, click, submit and boom,
here's a beautiful note thatyou can just be done with.
That is not the case.
Some ways you can bypass thatand help it are testing out some
(10:22):
of these different platforms.
Some of them will now let yougive it your own progress note
template and actually give itdirectives.
So you could say things like inthe intervention section
include three interventions.
Include five interventions.
Include five sentences.
In the summary section includea two-sentence summary.
(10:44):
In the summary section includea two-sentence summary.
So, as with anything with AI,the more instruction you give it
and the more specific thatinstruction is, the more helpful
it would be to you.
So again, it's not a matter ofjust log in and don't touch
anything and just expect thesystem to work for you
beautifully.
You have to work with it alittle bit and adapt it and kind
(11:04):
of know what you're looking for.
So it can be worth testing.
I will say we are all in areally special place right now
because these things are so new.
These developers the vastmajority of them who are in
charge of these AI companiesbecause I'm doing this YouTube
series and because I'm one ofthe few people talking about it
(11:27):
so openly they're all reachingout to me and they want to meet
with me and they want to dodemos and they want me to review
their platform.
So I've talked with most ofthem and many, with many of them
, have an ongoing relationship.
Most of them have really reallygood intentions.
They're really good people.
They have a lot of differentreasons for why they've started
these platforms.
They want a good product.
(11:49):
They know that there's highcompetition already and they
want their product to be reallygood and they love getting
feedback from their users.
So if you're using one or ifyou're testing an AI platform,
email them.
If you find it difficult, theywant you to reach out and say
I'm struggling with how to usethis or this isn't saving me any
(12:10):
time.
A lot of them, like the literalperson who helped create it or
is like in the back end, youknow, created it and is updating
things they will get on a callwith you and want to talk to you
and say what would you like itto do, what would be helpful.
So a call with you and want totalk to you and say what would
you like it to do, what would behelpful.
(12:31):
So use this opportunity,because in two years that is not
going to be the case.
You are not going to be able toget on the phone or on a Zoom
call with one of thesedevelopers.
So if you're struggling or ifyou just feel like I have one
but I'm not optimizing it, emailthem and say hey, it's still
taking me a long time to do mynote.
Do you have some tips for me?
(12:51):
Or I'd like to be able?
The summary is super long.
Most of them know they they'llstraight up be like okay, we've
seen your other reviews.
Ours is also really long.
We know this.
We're trying to work on it.
So ask them, can I give it adirective to only give me a two
sentence summary?
You know so.
So don't feel like you're stuckwith what it's giving you,
because I think some people login and get a little disappointed
.
It requires a little effort and, just like when you start using
(13:14):
an electronic health record.
It requires some getting usedto.
You're probably not going tolog in and immediately just hit
the ground running.
It takes some figuring out andsome testing and they're so
different.
So let me get back to we talkedabout.
You log into a session.
You use that platform to doyour telehealth session.
It records everything.
(13:34):
It gives you a transcript, itgives you a note.
There's a ton of variety allthe way to.
You log in and you type up asummary and then it creates a
note for you.
You can, for some of theseplatforms, you can, log in and
you type up a summary and thenit creates a note for you.
You can, for some of theseplatforms, you can, log in and
you can give it a verbal summaryof what happened in the session
and then it will create a notefor you.
For some of these platforms,you can check off that.
(13:58):
You don't want the transcript,it will just give you a note.
Let's see I'm trying to thinkso, yeah, you can do the
transcript, you can verbally doit, you can type in a summary,
you can upload a recording.
So if you're doing in session,in person sessions, then some of
them you can have that listenin on in person, or or you can
(14:20):
do that on a different type ofdevice and then have it listen
to that.
So there's a variety of optionsfor how you give it the
information.
Speaker 2 (14:29):
And yeah, my cat is that's all right, it's okay,
it's like those of us with kids,it's okay, thank you.
Yeah, my kids are launched, sohe may run in.
Yeah, I'm hoping she'd justcome and hang out, but okay,
yeah so questions about hippa,questions about confidentiality,
(14:52):
and got all these scary storiesabout, you know, people selling
information or they go out ofbusiness, and then where's the
information?
What so?
Speaker 3 (15:00):
talk a little bit about that yeah, yeah, so, and
these are kind of the things tolook out for.
So I'd say the first thing isthere's one ethics code that has
released a statement on usingAI, and I think that we can all
use that as a good example.
So, for example I'm apsychologist APA hasn't released
(15:20):
a statement.
They're not going to release astatement for probably years,
right, cause they wait to be thelast one.
They wait for everybody else tofigure it out, and then they'll
release a very vague statementabout what we're going to do for
AI.
But it's gosh, I always butcherthe name MHCA, lmhca, mhca the
Mental Health CounselorsAssociation.
So their statement essentiallysays that AI is a new technology
(15:46):
it's okay to use, as long asyou get informed consent, and
this kind of jives with what allof our ethics codes say in
general about anything that'snew.
Right, it's okay to trysomething new, to do something
new, as long as you are checkingthat it's as secure as you can
check for and as long as you areinforming people.
(16:09):
Hey, this is a new thing.
None of us can say that we knowwhat AI will be in two years,
right, and so the key is toinform people.
I would like to use AI to writemy progress notes, and, yes, I
would recommend so.
Again, apa has not released astatement, right?
So I have no, as a psychologist, no way to say I could get in
(16:33):
trouble for using or not usingit.
But I could tell you I canextrapolate from the rest of
informed consent policies andguidance on ethics and using new
technology that I better letpeople know this is new and that
I can't know all of thepotential downsides yet.
(16:53):
So that's one thing that'sreally key is to let people know
I'd like to use this AI andthen let people know are you
using it afterwards?
Is it not recording a session?
It's just something you want touse to write your progress note
, or is it something you want touse to record a session?
Because then that can get intostate specific things where
certain states say, if you'regoing to record a session, that
(17:14):
requires its own consent as wellas different ethics codes say
that too.
So so you might have a two partthing there where you're
talking about getting consentfor AI and then also getting
consent to record them.
So think about both of thosethings and, as you're hearing me
talk about this stuff, ifyou've already played around
with AI and you're like oh mygosh, I haven't done that.
(17:36):
Don't freak out, it's okay.
Just do it the next time yousee your clients right, that's
what we always say.
And just do it better from nowon.
And that's the thing with thisnew stuff.
You know, we will experimentand we'll learn and we'll learn.
Oh, I may have done that wrong.
Okay, let me do better.
And it's not anything that youneed to really freak out about
(17:57):
or worry about.
Speaker 2 (17:59):
So that's the thing that everybody signs a BAA.
Speaker 3 (18:03):
Yes.
So number one is informedconsent.
Right, make sure you getinformed consent from your
clients.
And I would say the other partof the MHCA guidance that I
think other ethics codes willprobably agree with is that if
your client says no, I don'twant you to use AI, you can't
then say I can't see you Like.
(18:26):
Ai is not a standard of carethat you can use to excuse not
using it or getting rid of aclient for not using it.
So if a client feelsuncomfortable with it, you have
to be OK with saying, all right,I won't use it.
So the next part is technologywise.
How can you know what is safe?
All you're required to do is tohave a BAA with that
(18:49):
organization.
So that's with pretty much anytechnology.
Ai is not different in thisregard.
Think of it just like an EHR,your electronic health record,
your electronic health record.
You are giving it very personalclient data.
You're giving it client records.
You're giving it clientinsurance numbers, date of birth
, everything right, and youdon't personally research that
(19:12):
technology to see if you feellike they are doing all of their
data security the right way,because we don't have degrees in
that.
We don't know how to do thatthat's the point of a BAA, a
business associate agreement.
That that's the point of a BAA,a business associate agreement.
That is a tech company, anytechnology, says we understand
that healthcare informationneeds to be more secure.
We understand that there isthis law called HIPAA.
(19:38):
We know it exists and we arecreating our technology to
coincide with that law and to bein compliance with that law.
And so we are giving you alegal agreement, a BAA, that
says we are protecting thisinformation in compliance with
HIPAA.
So all you need, as thepractitioner, is that BAA.
That's what makes you compliant.
Now, how you get the BAA is itvaries.
(19:59):
So for some of these companiesyou have to request it.
So you might be using it andnot have a BAA, and technically
you are not HIPAA compliant.
The software is still HIPAAcompliant, it's doing the same
thing.
They're not changing thesoftware as soon as you download
a BAA right, but you are notHIPAA compliant without that BAA
.
Some of them automatically justgive it to you, send it to you
(20:23):
as soon as you sign up.
Some of them require you tosign something when you sign up
and then it's downloaded.
So it's all different.
But you do want to check andmake sure that you have a copy.
It should be a physical copy ofthat BAA, and so you typically
get that during the sign upprocess.
(20:43):
But if you don't then reach outand ask for it and if they
can't give it to you, then youare not being HIPAA compliant if
you do not have that BAA.
And I emphasize that becausethere are companies
unfortunately there was one thatis now HIPAA compliant that for
a long time was not.
They were not giving peopleBAAs and they were saying they
(21:06):
were HIPAA compliant becausetechnically they're HIPAA
compliant.
But you are not HIPAA compliantif you don't have that BAA from
them.
So don't be duped.
That's the only thing you haveto do is get that BAA.
Speaker 2 (21:21):
Is there?
We have a.
I'm skipping around in thenotes because you're answering a
lot of this or not.
Notes the chat.
Is there any?
We have a.
I'm skipping around in thenotes because you're answering a
lot of this or not.
Notes the chat.
Is there any way you couldscreen share and just show us
what a note would look like fromAI?
I mean, you said you had somesample sessions or things that
you do.
Speaker 3 (21:40):
Yeah, let me pull up.
Let me pull it up before Iscreen share.
I gotta dig around through myyou know see all these handouts
that I have.
So I do have a handout Actuallyit's somewhere else.
Let me just search for it tofind it more quickly.
But I have a handout that hasall of the different progress
(22:00):
notes I have generated fromdifferent AI progress note
generators and what I use.
The session that I use is thesession with Carl Rogers and
Gloria.
If any of you are familiar withthat, it's a very popular
(22:22):
common training session from the60s and I'll go ahead and share
my screen if that's okay.
Then If any of you are familiarwith that, it's a very popular
common training session from the60s and I'll go ahead and share
my screen if that's OK, thenhold it up, all right, so, and
I'll share this link with youguys too, so you can reference
(22:43):
it.
This is a handout that I offerfor anybody as part of that
YouTube series anyways.
So I have some important noteshere and I have comments on all
these different platforms.
There are many platforms thatare not listed here yet.
I will say AutoNotes is onethat was not HIPAA compliant
(23:09):
previously and now is.
That still gives me a littlecaution that they at one point
were not and we're not quiteupfront about that.
So they're not my first that Iwould recommend.
But here's an example of atherapy session note, of adapt
note, data, assessment and plan.
As you can see, this is areally long note.
I mean, look at this, look atthe plan section.
(23:30):
Your plan section does not haveto be that long.
Okay, I'll just like give youlike the quickest note feedback
ever and say that, unless thisperson was suicidal and we
created a whole plan for what todo and they had never been
suicidal before, right, and sothis is like a really big deal
and we're making a wholedocumentation.
You know, we're changing ourtreatment plan and do all kinds
(23:52):
of stuff.
Your plan doesn't need to bethis long.
It could be one or twosentences really.
So here I emphasize in theorange these are kind of like
caution things where it madethings up.
So here's a good example, andthis is not to pick on auto
notes, but to just say all ofthese AI platforms do this.
It says moving forward, thetherapeutic plan will focus on
(24:15):
helping Gloria navigate herfeelings of guilt, et cetera, et
cetera.
There was no therapeutic plandiscussed.
This was a first and onlysession with this client.
Therapeutic plan, treatmentplans anything like that was not
a thing.
I don't think they even had todo that in the 60s, right?
So this was not discussed inany way.
But AI this is a general rule ofAI.
It wants to make you happy.
(24:35):
It wants to fulfill yourrequest.
This is a general rule of AI.
It wants to make you happy.
It wants to fulfill yourrequest.
And if it has been trained bythe developers to say it needs
to include certain interventionsor that always mention a
therapeutic plan, things likethat, it's going to do that,
even if that's not what actuallyhappened.
So it will create these kind ofclinical sounding things that
(24:58):
didn't exist.
And I want to emphasize that,because one of the big things I
hear from people when they startusing AI is like, oh my gosh, I
could never write a note thisgood.
And a lot of you might belooking at this and like, oh, I
want to write down what it saysor how it phrased it.
You don't even know if this isaccurate.
It could be totally makingstuff up.
You don't even know if this isaccurate.
It could be totally makingstuff up.
So this is not for me.
(25:22):
This is actually not new.
This has been going on and eversince I've been training people
on writing progress notes.
It's not about making a notesound clinical.
No one is grading you on that.
We want to see what actuallyhappened in the session, and I
would rather that a 12 year oldcould read your progress note
(25:44):
and understand what happenedthan for you to overcomplicate
it by putting in a bunch ofclinical jargon.
So it is much more importantthat progress notes are just
clearly read and understood.
Let me show you one that looksreally different.
So some of them, like I said,will give you only options like
this, where it's a big field oftext, and then typically what
(26:06):
you would do is you would copyit all and paste it into your
EHR or whatever record systemyou have, and so you might tweak
it, adjust it, only copy pasteother parts over, et cetera.
Actually, I will highlight herebecause I'm going to highlight
this in the next note too whereit says Gloria is currently
facing sorry that doesn't makeit easier to read, does it
(26:29):
Significant emotional distressprimarily centered around her
feelings of guilt and fear ofrejection.
Her mood was anxious and hertearful affect suggests a high
level of emotional turmoil, etcetera.
If you have seen this session,I would not say that Gloria
exhibited significant emotionaldistress or that someone being
(26:52):
tearful suggests a high level ofemotional turmoil, right, would
we say that about someone who'scrying because their father
passed away and they got tearful?
No, that's normal, right.
And so I will say that some ofthe AI platforms because they
are trying to incorporate whatthey know we need, which is
(27:16):
often to show medical necessityit does tend to over-pathologize
or overemphasize things andit'll take what might be a
stressor or a worry or a concernand it then blows it up to be
like significant distress.
And there is a balance of yes,we do want to show things and
(27:37):
document well, for insurance,but I wouldn't want someone in
this session she primarily istalking about she lied to her
daughter about having sex, andher daughter is nine, I think,
and so I think we could.
There are different people whocould argue that, like, that is
totally appropriate to lie toyour daughter about having sex
(27:59):
when your daughter is nine yearsold, right?
And then there are people whocould argue well, it actually
could be a really healthyconversation to have.
All of those could be right,answers right.
The point is that's a reallynormal parenting experience and
she was distressed about it.
She's more distressed about,like, essentially, all the
(28:20):
things going on.
Am I going to mess up my kid?
Is it going to mess up my kidIf I tell her I lied?
I feel guilty, but is it, youknow?
Should I just deal with that onmy own?
These are normal parentingexperiences that people have,
right, um?
And so I worry that some ofthese notes will overemphasize
things like that and just kindof inaccurately reflect
(28:44):
someone's actual level ofdistress.
So I wanted to mention that.
Let's look at.
Mentalic is another one that Iwill say they win shortest note
award.
So although it looks long, youcan see it's like broken up a
(29:06):
little bit better and they don'thave giant paragraphs of things
.
Um, it is written very similarLike.
It looks more like a psychiatrynote than a psychotherapy note
to me, but it still has prettygood.
Here it said client isstruggling with adjustment to
single life after divorce, guiltover lying to her daughter,
concerns about her relationshipwith her partner.
(29:27):
There's no partner, there's noconcern about relationship with
a partner.
They just made this up.
Single life after divorce, yes,but no, that's not there.
So there are times where, again, I don't know if the AI took
her talking about her divorce tokind of mean partner, how
(29:48):
that's working.
But these companies, you haveto read these notes and you have
to edit them and unfortunately,you have to read them in a.
You have to edit them andunfortunately, you have to read
them in a fairly detailed way inorder to know if it's accurate
or not Right.
So, like this one phrase Iwould take out.
The rest of it is fine, butlet's just browse through this
(30:12):
note and then, like I said, I'llgive you guys this, this
handout, and then this handoutis the one that I update every
time I do a new review, soyou'll continue to get those
updates if you follow along withthat YouTube series.
(30:32):
Then interventions it kind ofbreaks things out into approach
and modality, actualinterventions here I'll pick on
mentalic because it's still apretty good note.
Here it's kind of doing theopposite.
The therapist used activelistening, reflection,
validation and empathicresponses to help the client
feel heard and understood.
This is Carl Rogers.
100%.
This is how Carl Rogers works.
(30:53):
This is an accurate statement.
This is not what an insurancecompany wants to see for your
therapeutic interventions, right?
So here we're constantlybalancing these two things, and
this is why I call documentationan art.
This is constantly the art thatwe're doing is balancing both
(31:14):
insurance type statements,progress, what insurance wants
to see, and what actuallyhappened.
Now I think that we can alsoidentify other interventions
that Carl Rogers presented thatare more insurance appropriate.
Here I'll give you a littlesidebar tip.
So how do you know if yourinterventions are appropriate
(31:39):
for insurance or not?
I like to use what I call thegrandma rule.
So if you, assuming you have agood relationship with a you
know a nice grandmother, withyour nice grandmother, how would
you expect her to listen to you, or how do you expect
interactions with her to go?
Are you expecting her toactively listen?
(31:59):
Probably.
Are you expecting her to beempathic?
Probably.
You can also extrapolate thisto your best friend.
How would you talk to your bestfriend about something?
Are they validating?
Yes, you don't need a graduatedegree to do any of those things
, right?
Do we hone those skills ingraduate school?
(32:20):
Absolutely.
Are those necessary foreffective therapy?
Absolutely, but those arebaseline things that you don't
need a graduate degree to do.
So for insurance.
That's why I call it thegrandma rule.
What is something that you hadto go to graduate school to
learn to do?
Those are the types ofinterventions that you want to
(32:42):
focus on having in your progressnotes.
So here, mentalic kind ofover-pathologized the client and
then kind of underrepresentedwhat the therapist did from an
intervention response.
But I will say, with a lot ofthese things you can teach these
platforms what you want it tosay.
(33:03):
So there is one Clinical Notes,ai, where you can give it a lot
more direction and you can tellit, for example, choose five of
these interventions to put inmy progress note and you could
give it a list of 50interventions and that way you
know that at least it's fromkind of an approved list that
you've curated.
(33:23):
All right, let me stop sharingand just check in here and see,
see how everybody's doing andwhat you guys think.
Speaker 2 (33:31):
Well, one of the questions I saw.
But so in, I think this is.
But so and I think this isdovetailing on what you're
saying it's not really able.
Like with ChatGPT, we could putin something we had written and
ChatGPT will start giving usresponses that sound like us.
So this isn't like that.
We're not able to train it thatwell yet.
Speaker 3 (33:54):
It depends on the platform.
So this is why I need tore-review.
So, for example, clinical NotesAI.
I think you can do that alittle bit more.
Session Aid is another one, butthe way you're going to do that
really well is by using aplatform where you can give it
(34:14):
those really specific directivesfrom the beginning and then
maybe adjusting your templatebased on how it's writing.
I do think Orchid is one that'san EHR and an AI platform and
(34:38):
that one does learn how you workover time because it has a lot
more information from you,because it's literally
integrated within the EHR, whichis likely the future of how
this will work, and so I dothink that is what this will
look like in two to five years.
But right now it's a lot moreeffort and just like if you
found with ChatGPT, you stillhave to do a lot of work to oh,
I forgot to tell it this, oh, itgave me this and I realized I
(35:00):
actually want somethingdifferent, and then you have to
rework it.
So you're kind of doing thatwith your template and with the
directives you're giving the AI.
Speaker 2 (35:12):
Okay, One question was we hear a lot of scary
information about AI taking theplace of clinicians in the long
term.
Speaker 3 (35:19):
What are your thoughts on that?
All you have to do is readthese notes to know that's not
happening anytime soon.
It's not.
I mean, I train people on howto write progress notes.
I'm not worried about AI takingmy job, so it's definitely not
taking your job.
I do think, and why I'm excitedfor this next round of the
(35:41):
reviews I'm doing in the nextmonth.
A lot of the companies aregetting on board with kind of
this integrated approach of theclinician.
Can, you know, use the AI towrite progress notes and send
messages to the client, sendhomework assignments to the
client, and the client can login too and check up on things.
(36:03):
So I know that young sidekickis one that I'm about to test
that offers that already.
Ooh, and I'm going to blank onthe other one.
So I feel bad, but they'll allget YouTube videos so you'll be
able to see those.
But I'm really excited aboutthat because I think that's the
reality of how we can use theseand how clients will be able to
(36:24):
use them.
So think about, in one of thegroups that I run, people were
talking about some people weretalking about the benefits of
doing sessions every other week,for example, and one therapist
I work with was saying they workwith a lot of trauma and a lot
of times, sessions are reallydeep and intense, and so they
actually find it more beneficialto take a break and do sessions
(36:47):
every other week.
Well, if you could add on an AIcomponent, where then the AI
knows to like follow up with aclient, maybe a week later, or
give an assignment, and thenthey can log in and have some
kind of an app that they canreference To me, that would be
hugely beneficial to thetherapeutic process and I think
(37:08):
a lot of people would like that.
So I'm excited for that.
And where it's an integrated,clinician can use it for these
things and client can use it forthese things, and client can
use it for these things.
Related to that, I do want tobring up one other caveat that I
think is really important andhasn't been brought up as much,
which is, as I'm talking aboutthis, think of all this data
(37:31):
that's going back and forth inthis AI platform and this is the
AI platform.
In this AI platform and this isthe AI platform.
It's not your EHR and it hasmaybe it has a full transcript
of every session.
It has maybe client reportsthat they're submitting to you.
It has now also progress notes,maybe three versions of the
(37:52):
same progress note that yourevised or that you had it
produced for you in differenttemplates that you revised or
that you had it produced for youin different templates, and so
now you essentially have clientPHI in two systems and you're
maintaining it in two systemsand they could be two HIPAA
compliant systems.
That's great.
(38:12):
What happens when that clientsubmits a records request and
they say I'd like a copy of myrecords and you give them a copy
of the records and they saywell, what about the transcripts
?
Isn't it giving you atranscript?
Isn't it producing a transcriptof the whole session?
I worry that that is going tohappen one day to an
(38:35):
unsuspecting therapist who willthen realize oh like, releasing
a whole transcript to someone isa little different than
releasing a progress note andit's their record.
And in the US and in Canada aswell, it's really difficult to
(38:56):
tell someone you can't accessyour own record.
So think about if you went to aphysician and there was a
transcript of that session andyou knew there was a transcript
and you said I'd like a copy ofthat and the physician said no.
That now becomes a client rightsissue, right?
I think that the law in mostcases and in most states would
(39:18):
side with.
People get access to theirmedical records and personally I
think that's probably the rightcall.
Right, people need access,should have access, to their
medical records.
So what does that mean?
For us, that means, if you wantto use one of these platforms,
that that is a question to askthem.
When you sign up, do youautomatically delete these
(39:41):
things?
So some of the platforms willlet you create that as a setting
.
So at once a week it's deletingall the notes or it's deleting
all the records.
Some of them will let youoverride that.
Some of them automaticallydelete that.
So session aid, I think, is onethat is very transparent about.
We do not plan to be an EHR.
We do not plan to maintain thisinformation long term.
(40:04):
It's going to be regularlydeleted.
That could be really beneficialfor managing this issue.
It could also be a realstruggle if you're using it
because you tend to get behindin your progress notes and you
want to have access to thoseprogress notes two weeks later,
right?
So that's why I'm bringing itup, as it's not necessarily a
(40:24):
good or bad thing, but you doneed to consider that ahead of
time and what you're going to dowith it and what the platform
offers you in the way of optionsfor deleting or saving the
information options for deletingor saving the information.
Speaker 2 (40:44):
So one of the questions here would love to
just know how we can bestutilize AI in an ethical way,
then mostly to save time withtreatment plans, you know.
Is there like a hybrid way touse it, Like if we're not just
recording everything, then howwould you use it?
Or how would you, how do yourecommend that we use it?
Speaker 3 (40:59):
Yeah, I think this is where I don't think it's always
the solution and it reallydepends on what your particular
problem with progress notes is.
If it's just time, it's rarelyjust time, I will say.
Usually the time issue isactually it takes me 20 minutes
to write a progress note becauseI sit there and question
everything I'm writing andrewrite it.
(41:20):
So that's time is not yourissue.
It's feeling confident in whatyou're writing.
That's actually the problem tosort out there, right, and I
would say that's the most commonscenario I see.
And if time is your issue, isit that you're ending?
Are you ending sessions on time?
That's not going to change withAI, right?
(41:40):
Are you giving yourself enoughtime to write progress notes?
Do you have time in yourschedule for progress notes?
You're still going to need thatwith the AI platform because
you'll have to review that,review them and then copy paste
it into your EHR, right?
So, really, you know, I dothink it's the first thing I
always recommend for everyone islook at what is the real reason
(42:02):
that I'm struggling withprogress notes and try to
address that first.
If it is genuinely time, whichI do understand, that is a big
part of it for many people.
That is where this hybridmethod can be really helpful.
That is a big part of it formany people.
That is where this hybridmethod can be really helpful.
But for me, the hybrid method,I think, is more effective if
(42:23):
you are literally thinking of itlike collaborative
documentation.
So for treatment planning, forexample, I always recommend you
never need to do a treatmentplan outside of a session.
The treatment plan is literallywhat you and the client are
agreeing to do in therapy.
There is nothing more valuablethat you could talk about with
(42:44):
them, right, you want to?
Literally.
It's an opportunity toliterally be on the same page
about what you can accomplishtogether, what they can expect
from you, what you can expectfrom them, where you want to be,
where you want to work andfocus on.
I don't think that you need tosit there and you know, like
(43:05):
some of us have done in agencydays, and spend half an hour
filling out a really long formwith them, but absolutely
talking to them about the goalsthat you want to work on and
identify like why are you here?
What is it that you would liketo work on?
There is no question, based onresearch, that that is a
(43:29):
valuable clinical discussion tohave, and then having that
clinical discussion will have apositive impact on the therapy
you do with clients and it'sempowering for people.
So to me, that is something youdon't really need AI to help
you with.
Now, where that might come up isyou say, okay, but how am I
supposed to sit here and writethis with a client when I'm
thinking of what to write andI'm trying to produce this
(43:51):
content and, yes, you're saying,make it understandable, but I
still got to write anintervention, right, and it
takes me some time to think ofthat.
And so that's where I thinkthat having cheat sheets and
templates and I think that wasone of the questions that came
in ahead of time too, that youhad sent me that having like
(44:12):
pre-created statements is themost valuable thing.
And that way you're just eitherfilling in the blank or you're
circling things or you'rechecking off boxes, but they're
detailed, they're not.
When I say checking off boxes,I don't mean, like what your EHR
gives you as the standardcheckbox interventions list.
That says cognitive behavioraltherapy that's a thousand
(44:34):
interventions, right.
We're not checking that off asan intervention.
Cognitive restructuring that's amore specific intervention.
Creating a pros and cons listintervention, creating a pros
and cons list.
That's a more specificintervention.
(44:56):
You know, reframing, groundingwork, deep breathing exercises,
exploring patterns in familybehaviors, exploring patterns in
the family.
I would say Carl Rogers didthat in that session with Gloria
, but that's not one of theinterventions that any of these
AIs are picking up from thatsession.
But he absolutely did that andtalked about how her
relationship with her father ishaving an impact on this
(45:19):
situation she's having with herdaughter, and so it was a very
active reflection.
So, to me, curating a list ofstatements that you use and of
phrases that you use that applyto the way you work and the
clients you work with is themost useful thing.
And if you can use that to thengive one of these AI platforms
(45:43):
a template and a cheat sheet,then it's already going to sound
like you and that's kind ofbypassing what you're talking
about with, like using chat, gptand having it teach you how to
write.
So using one of the platformswhere you can give it a lot of
direction and you can reallycurate the template to your own,
I think is the best bet.
Speaker 2 (46:08):
Well, and I think that helps with the next
question as well, because whenyou talked earlier about the
quality of the note and it, youknow you talked about, you know
making up things.
Well, if we're trying to alsocreate a note that protects our
client, right, because of thethings like we're in Texas and
there are things that ourclients are going to discuss and
we don't want to put that inthe note, and so I'm hearing you
(46:33):
say that a template with thesesort of responses that you
usually use could really behelpful.
Speaker 3 (46:40):
Yeah, and I always recommend that too, when I've
done a couple of YouTube livesas well on this particular topic
, like, how do you documentsensitive topics and this has
become much more of an issueover the last few years right,
and so having I think, whetheror not you use AI having some of
(47:01):
these statements ready to go sothat you don't have to think
about them in the moment whenit's stressful, is really
helpful.
So I'll give you some reallyspecific examples.
So, let's say, your client istalking to you about their
feelings around having anabortion and, regardless of the
state you live in, that could bea really heavy and important
(47:25):
topic to your client and is areally emotional experience,
regardless, right or could be,and so that's pregnancy loss,
and pregnancy loss allows you tocapture the emotion around the
event without actually talkingabout the event itself and how
it happened, actually talkingabout the event itself and how
(47:49):
it happened.
But having some kind of likepre-created statements like that
for these highly sensitivetopics will help, and I
encourage people you know youcan watch one of my YouTube
lives to get information on thatto get ideas for what to write.
Talk with your colleagues aboutwhat they write, but the fact
is, if you have a client who'sreally stressed about that
experience, it might besomething they're talking about
for months, right?
(48:09):
So it's not a matter of like.
Let me be as vague as possibleand never document this.
That might be an extremely thatis an extremely relevant thing
to document for many clients,and so having an idea of how to
phrase it is really, reallyimportant.
Speaker 1 (48:28):
Excellent.
Speaker 2 (48:31):
So, if it's okay with you folks, if you have the hand
, raise emoji or if you want topost in the chat, I'd love to
open it up to your questions Ifyou'd like more detail about
things that Dr McCaffrey'stalked about or just more
information.
So Alyssa says how to phrase itverbally as it is recording or
(48:55):
in the edits.
Now, alyssa, if you unmute,you're going to be on the
recording.
Speaker 1 (49:01):
So I totally get it if you don't want to do that,
but I'm fine with that.
Okay, I don't care.
My concern is so.
I'm just like I hear all theHIPAA and I hear the business
associates, agreements and allthese kinds of things, but in
the end of the day, someone islistening into my session,
whether it's AI or not, and thenthat's being transcribed.
So when you said phrasing, areyou referring to as we're
talking to them to change ourdialogue?
(49:21):
Are you referring to when weget the note that we have to
edit and everything that we'rehaving go-to statements for that
?
I was just unsure about thatpart.
Speaker 3 (49:29):
Yeah, really good question.
So that might be aconsideration for whether or not
you use recording, Because tofollow up with that, you know
very real example of a clientdiscussing concerns around an
abortion, right?
Maybe they're deciding what todo.
Whatever that might be, Oncethat word pops in, it becomes a
(49:49):
very charged issue.
So is your AI platform going topick up that word in a
transcript?
Probably and that may or maynot be something you can that's
likely not something you cancontrol, so that's something
something to think about.
Now, that's absolutely.
You can delete the transcriptimmediately or make sure that
(50:13):
you're using a platform whereyou can, right, Because maybe
your practice is just you deletethe transcript immediately for
all sessions, not a big deal.
But the AI needs some data togive you a progress note, right,
and the more data it has, thebetter note it will give you.
Just like with anything, Icould give you a better progress
(50:34):
note if I have more data aboutthe session too, right?
So anything that's, it's alwaysgoing to work that way.
So if you, you could either notdo the recording and you could
curate the information you giveit, and that way you can just
straight up use the wordpregnancy loss from the
beginning.
It never hears anything else,and that's what the AI will use.
Or you can wait for it to giveyou the information and then you
(50:57):
edit it in the note, but thenyou would want to make sure that
you delete any transcript thatthere is.
And I will say you know, andhaving a transcript isn't always
all bad.
If you do supervision, thatcould be awesome, because you
get way better information aboutwhat's happening in sessions
(51:17):
than you will secondhand fromtalking to trainees, right?
So you know, it's not that it'sright or wrong, but it's
important to think through allthe potential ramifications
Awesome, Thank you.
Speaker 2 (51:33):
Yeah, Keisha wants to know is there one platform that
you found to be the mostaccurate in capturing the
session without making things up?
And you said I know, you saidyou're re-reviewing these things
, but what have you found?
Speaker 3 (51:46):
Yeah, no, I haven't really.
They all like they all madesome kind of mistake at some
point.
I forget which one.
Like one of them, even this isa good example of how AI is not
going to take your job.
Okay, so one of them literallyphrased something how Carl
Rogers did.
(52:08):
But guess what Carl Rogers did?
He reflected back to Gloriawhat she had said and she had
said I forget exactly the phraseit was, but I would find it if
I could remember which platformit was but she specifically said
something like you know, mydaughter's going to hate me if I
do this.
And he said you're afraid yourdaughter's going to hate you if
you do this.
Right, he just reflected itback to her, which is what a lot
(52:32):
of us would do.
And so then the the, the noteit generated was like you know,
therapist told client that herdaughter would hate her if she
did this.
So it took it out of context,right, so in context, like I
never would have even thought tophrase it that way.
But then the AI, it's just anAI, it doesn't know.
So things like that are anexample of how it didn't get it
(52:52):
wrong.
It wasn't necessarily wrong,but it was out of context, and
so those are the things that youhave to look out for.
Speaker 2 (53:01):
There was a question earlier about shredding a note
once you put it into AI, and Iknow we've talked about deleting
notes and things like that.
I want to be really clear.
There is nothing in the LPC,lmft, bhec, social work.
Well, actually I take that back.
Lmft speaks to how we get ridof our notes, or at least it did
(53:22):
.
I don't know if they do anymore.
So this is all best practicesat this point, right?
Until we have more precedent,until we have things written
into the rules.
So this is the best we've gotright now, right, right, melissa
?
I mean.
Speaker 3 (53:38):
Yeah, and I would also say I think it can be
helpful because a lot of thisstuff is so vague and this is
literally why I have a job,because documentation
requirements are vague so it canbe really helpful to make it
less vague by like what is yourprogress note?
And really define that.
What is your progress note?
If you use an EHR, it'stypically whatever note is
submitted to that EHR.
(53:59):
So if that's my progress note,any other notes I have floating
around need to be gone.
So if that's a transcriptsomewhere, if that's the
progress note, that's in the AIplatform.
That is unedited, because then Iput it over here and edited it
and maybe I edited it a littlebit over there, but then when I
copy pasted it, oh, I didn'trealize that and so I made a
(54:20):
change over here.
So now I have two differentversions of the same note.
Right, I mean, that wouldabsolutely happen.
It could get very confusing.
The easiest way is just once Ihave that progress note in that
system, I get rid of all myother notes, and so you could
also shred.
Yeah, if you took notes, if youtake tons of people take notes
by hand, totally fine, you canabsolutely do that.
Take your notes by hand andthen, as soon as it's in an EHR,
(54:43):
shred that.
Or in your computer somewhereelse, shred it.
So kind of make a rule foryourself about where you're
going to have your notes.
What is the note, and just keepthat one, okay.
Speaker 2 (54:57):
We're getting requests here if you have sample
informed consent.
So this is a good time, I think, for you to drop some links
into the chat and everybody.
You know the drill when you seea link, click it, Because once
the meeting ends, the links goaway and then you'll have that
tab open amongst all thehundreds of other tabs I'm sure
you have open.
What's the best way to contactyou?
Speaker 3 (55:21):
Yeah, so the best way , I will put in the link for my
website and then I will will putin the link for my website, um,
and then I will also put in, um, the direct link to the book as
well.
So, um, my business called QAprep.
Here's the website, qaprepcomfor those listening later, so
(55:42):
just qaprepcom.
Uh, takes you there.
The book is called Stress-FreeDocumentation for Mental Health
Therapists.
There is a paperback versionand a Kindle version and Amazon.
It's available on Amazon.
Amazon randomly puts things onsale and it is on sale right now
(56:06):
and I still get the same amountof money.
So go buy it on sale.
And it is on sale right now andI still get the same amount of
money.
So go buy it on sale.
Um, yeah, um, and that book has.
I'm going to put the name ofthe book in the chat stress-free
documentation for mental healththerapists um you've got a
follow-up.
Speaker 2 (56:24):
Has been following your your youtube chin page oh
exciting.
Speaker 3 (56:29):
Thank you, keisha.
Yeah, and in in the book it hasall of my templates, it has my
full paperwork packet, and sothere's also links in the book
so that you can log in and getthem electronically.
So you don't have to sit thereand photocopy or copy copy into
your own.
You can just download the formsand and use them directly.
(56:49):
So that'd be the book I wouldsay is the easiest way to get a
copy of of my all my consentforms.
Speaker 2 (56:56):
Excellent.
I love that, thank you, thankyou.
Thank you so much, andeverybody.
This has been wonderful.
I'm so glad you're here.
Step it uppers.
If you are in the membership,you will get this as an
asynchronous CE.
Once we process it, y'all knowwhat to do.
Go visit Dr McCaffrey, get yournotes in shape, get rid of the
(57:17):
get rid of the shame of gettingbehind that I know nothing about
.
And, yeah, enjoy your Mondayevening.
I think you made us all feelbetter.
Dr McCaffrey, thank you so much.
Speaker 3 (57:28):
Good, good, I'm glad.
Thank you all for yourquestions.
That helps so, so much and Ireally I value your trust and
vulnerability there.
So thank you all.
Speaker 2 (57:38):
All right, we'll see you later.
Guys, have a wonderful evening.
Bye.
word and uh.
Welcome dr melissa mcafree andwelcome to uh texas counselors.
Creating badass businesses freeworkshop awesome, thank you.
Speaker 3 (00:18):
Thank you for having me, and you will have to um
excuse the unprofessionalpresentation today, meaning that
my cat went to the vet thismorning and was medicated and is
just kind of loopy and whining.
So she may make an appearanceand you may hear her, and I
apologize.
Speaker 2 (00:38):
That is totally okay.
Speaker 3 (00:40):
Thanks, hopefully you can't hear her in the
background right now.
No, I don't hear her.
Kat, you're good?
Okay, good, good.
But yeah, I saw that we had afew questions coming in and yeah
, so I'll let you take it fromthere and where we should go
next.
Speaker 2 (00:59):
Excellent.
Well, I just made you co-host.
So if you need to share screenor anything, let us know
Everyone else.
You know-host.
So if you need to share screenor anything, let us know
Everyone else.
You know the drill.
If you need CEs for thispresentation, leave your camera
on and we will post a link.
Oh, we just did Okay.
So there's the link.
It looks like I'm posting it,but it's actually my assistant,
jennifer, who's logged in as me,because you know wifi.
So here we go.
Speaker 3 (01:21):
Oh, awesome, and I I don't have a presentation, so I
thought this was more just a Q&A.
Oh no, no, absolutely.
So if we wanted to start outwith those, then that would be
great.
Or if you have anything inparticular you want me to kind
of riff on, I can talk forever,so please interrupt me.
(01:41):
I love this talk forever.
Speaker 2 (01:42):
So please interrupt me.
No, no, this is great, so Iwill.
I will ask you guys take 30seconds or less and put it to
the chat.
Why are you here?
Why are you here?
What are you hoping to learnabout documentation?
What is worrying you?
What is?
Does AI enter your brain?
(02:04):
Are you looking to AI assomething that could save the
day, or are you looking at it asTerminator 5 or whatever we
think AI might be?
So start posting.
Speaker 3 (02:16):
Let me see.
And yeah, and I could say thatI would say there are some few
common things that people reachout to me.
For Obviously, we can get asnitty gritty and in detail as
you want and we do sometimes butthe most common things that I
hear from people is like how doI balance?
How much is too much in aprogress note?
(02:36):
When am I writing too much,when am I writing too little?
Related to that, one thatpeople don't often publicly say
but is an extremely commonexperience, is how do I catch up
on my progress notes, because Ifell really behind.
Speaker 2 (02:52):
I don't know anything about that.
Speaker 3 (02:53):
Yeah, and I can't tell you how many people have
told me.
You know, please, you know Ihave to be the worst one that
you've seen.
You know, like there's no wayyou've heard how you know that
someone else is as behind as Iam and I'm like Nope, I have, I
promise you.
So I've met people who havenever done notes so hard to beat
(03:13):
that right.
So you know, it's such a commonconcern, though, and it's a
very shameful experience to haveand to admit, on top of the
whole experience of just notfeeling confident in
documentation and both of thosein my like anecdotal, my own
experience with working withpeople and I've been doing this
(03:34):
for 10 years now has nothing todo with the school you went to,
with the license type you have,with the amount of experience
you have.
I work with people who have beendoing this for decades and
people who are just fresh out ofgrad school.
So those are two really commonthings, and I would say that,
(03:54):
because you say we want to getinto the AI stuff, ai is a
problem or is a solution thatpeople are hoping for that
people think will solve eitherof those problems, right?
So AI will write my note betterthan I can.
Ai will choose interventionsthat I can't think up in my own
(04:15):
head.
Ai will write it in a clinicalway.
Ai will include the informationthat insurance wants to see, or
, you know, ai will help me dothis faster, and so I think of
those two really common issuesthat I work with people on
specifically for progress notes.
I think that's what theyusually want AI to do, and I
(04:38):
will be the bearer of both badand good news in that.
I have worked with people whohave said AI is like they've
used the phrase life changing,so it really can help, and you
still have to know what goes ina progress note to be able to
use AI, because let's startthere then.
Speaker 2 (05:01):
Yeah, I mean, like the first, I'm looking.
We kind of released.
You know we have tons ofquestions now, but I see the
first one.
I see HIPAA security and I knowyou'll talk about that, but
wanting to hear how AI canassist with clinical
documentation and you canaddress both of those things
right, the quality of the noteand can it help me catch up?
(05:22):
Right, those two things yousaid are kind of the most, and
this would be a good time tomention too.
You help people with this andyou just wrote a book, so you
have lots of resources for usand you will be put up there.
It is All right good, and youwill be putting the link in the
chat for us before we end todayas well.
So, yeah, talk about AI and howit can help with those two
(05:44):
things, or if it can, or what weneed to do to make it do that.
Speaker 3 (05:47):
Yeah, absolutely yeah , and I think we can combine
quite a few of these questionsinto a kind of long winded
answer.
But please continue puttingyour questions in the chat.
It is helpful if you putquestion and then a colon before
you write your question out, asfar as being able to see like
who's just following up onsomething, who's writing a
comment and who is actuallyasking a new question.
(06:08):
So for AI, let me back up.
So, yeah, we'll kind of talkabout how it can be useful, what
to look out for, and when wetalk about what to look out for,
we'll talk about HIPAA.
So, because it's actually a lotmore simple than you'd think,
that part of it.
So with AI, let me talk abouthow some of these work.
(06:33):
I think that's a good place tostart.
So there are various options.
You can log in all the way from.
The AI platform will offer likea telehealth platform and you
will do your session overtelehealth that way and it is
literally recording the entiresession, listening into the
whole session on both sides, andthen after you end your session
(06:57):
, it creates typically atranscript, because it's
listened to the whole thing, anda progress note for that
session.
That can sound both veryawesome and very scary for many
therapists.
So you know and both feelingsare relevant this I have so many
(07:19):
things because I do like awhole AI training on this.
So with that, I would say, mybiggest takeaway for anything AI
.
So, regardless of what I'mgoing to say in any of this
other stuff, and if you takeaway anything, you have to edit
the AI note.
You must.
You have to know what should bein that note.
(07:42):
So I'll give you some examples.
I use the same therapy sessionto test a lot of these AI
platforms and I have a freeYouTube series where I'm testing
all the platforms.
I had to take a break while Iwas finishing up my book, so I'm
just now getting back intotesting more of them and needing
to retest them because theychange so frequently that now, a
(08:05):
year later, some of my reviewsare almost obsolete because they
have new features.
Every single one of them hasmade up information, every
single one has gotten somethingwrong and I would say my biggest
critique of overall, all the AIplatforms is that they are
extremely wordy.
(08:26):
They give you these huge giantsummaries of information and so
time-wise for most people.
Ai doesn't really save timebecause you have to the time to
read through the mountain oftext that it gives you and then
either delete it or even justsay like it's perfect as it is,
(08:47):
which I have never seen, I willtell you.
But let's say you only had tochange one thing.
It really does take a lot oftime.
Now that doesn't mean it's notworth trying out if you really
struggle with notes, because ifit saves you the mental effort
sometimes, if it takes the sameamount of time but you are not
needing to sit there and producesorry, if you can hear my cat,
(09:10):
she's being very loud If you'renot needing to produce all of
that information from your head,that can be a huge lifesaver.
I mean, that can be dramaticand it might mean the difference
between you avoiding writing anote and you feeling comfortable
logging in and getting it doneif it doesn't feel like it'll be
a lot of work.
So I wouldn't say that's a.
(09:33):
It's not necessarily a drawback, it just is a fact.
The other thing is that most ofthese AI platforms are much
slower.
If you have used something likeChatGPT for other things,
that's not how this is going towork.
You're going to hit submit oryou're going to end your session
and you're going to literallylike they'll even say, go grab a
coffee, enjoy doing somethingelse while we write your notes.
(09:56):
So I mean, it takes a while forthese systems to produce a
progress note for you.
So you know, especiallyespecially if they're processing
like an hours long session.
So it's not like end yoursession, click, submit and boom,
here's a beautiful note thatyou can just be done with.
That is not the case.
Some ways you can bypass thatand help it are testing out some
(10:22):
of these different platforms.
Some of them will now let yougive it your own progress note
template and actually give itdirectives.
So you could say things like inthe intervention section
include three interventions.
Include five interventions.
Include five sentences.
In the summary section includea two-sentence summary.
(10:44):
In the summary section includea two-sentence summary.
So, as with anything with AI,the more instruction you give it
and the more specific thatinstruction is, the more helpful
it would be to you.
So again, it's not a matter ofjust log in and don't touch
anything and just expect thesystem to work for you
beautifully.
You have to work with it alittle bit and adapt it and kind
(11:04):
of know what you're looking for.
So it can be worth testing.
I will say we are all in areally special place right now
because these things are so new.
These developers the vastmajority of them who are in
charge of these AI companiesbecause I'm doing this YouTube
series and because I'm one ofthe few people talking about it
(11:27):
so openly they're all reachingout to me and they want to meet
with me and they want to dodemos and they want me to review
their platform.
So I've talked with most ofthem and many, with many of them
, have an ongoing relationship.
Most of them have really reallygood intentions.
They're really good people.
They have a lot of differentreasons for why they've started
these platforms.
They want a good product.
(11:49):
They know that there's highcompetition already and they
want their product to be reallygood and they love getting
feedback from their users.
So if you're using one or ifyou're testing an AI platform,
email them.
If you find it difficult, theywant you to reach out and say
I'm struggling with how to usethis or this isn't saving me any
(12:10):
time.
A lot of them, like the literalperson who helped create it or
is like in the back end, youknow, created it and is updating
things they will get on a callwith you and want to talk to you
and say what would you like itto do, what would be helpful.
So a call with you and want totalk to you and say what would
you like it to do, what would behelpful.
(12:31):
So use this opportunity,because in two years that is not
going to be the case.
You are not going to be able toget on the phone or on a Zoom
call with one of thesedevelopers.
So if you're struggling or ifyou just feel like I have one
but I'm not optimizing it, emailthem and say hey, it's still
taking me a long time to do mynote.
Do you have some tips for me?
(12:51):
Or I'd like to be able?
The summary is super long.
Most of them know they they'llstraight up be like okay, we've
seen your other reviews.
Ours is also really long.
We know this.
We're trying to work on it.
So ask them, can I give it adirective to only give me a two
sentence summary?
You know so.
So don't feel like you're stuckwith what it's giving you,
because I think some people login and get a little disappointed
.
It requires a little effort and, just like when you start using
(13:14):
an electronic health record.
It requires some getting usedto.
You're probably not going tolog in and immediately just hit
the ground running.
It takes some figuring out andsome testing and they're so
different.
So let me get back to we talkedabout.
You log into a session.
You use that platform to doyour telehealth session.
It records everything.
(13:34):
It gives you a transcript, itgives you a note.
There's a ton of variety allthe way to.
You log in and you type up asummary and then it creates a
note for you.
You can, for some of theseplatforms, you can, log in and
you type up a summary and thenit creates a note for you.
You can, for some of theseplatforms, you can, log in and
you can give it a verbal summaryof what happened in the session
and then it will create a notefor you.
For some of these platforms,you can check off that.
(13:58):
You don't want the transcript,it will just give you a note.
Let's see I'm trying to thinkso, yeah, you can do the
transcript, you can verbally doit, you can type in a summary,
you can upload a recording.
So if you're doing in session,in person sessions, then some of
them you can have that listenin on in person, or or you can
(14:20):
do that on a different type ofdevice and then have it listen
to that.
So there's a variety of optionsfor how you give it the
information.
Speaker 2 (14:29):
And yeah, my cat is that's all right, it's okay,
it's like those of us with kids,it's okay, thank you.
Yeah, my kids are launched, sohe may run in.
Yeah, I'm hoping she'd justcome and hang out, but okay,
yeah so questions about hippa,questions about confidentiality,
(14:52):
and got all these scary storiesabout, you know, people selling
information or they go out ofbusiness, and then where's the
information?
What so?
Speaker 3 (15:00):
talk a little bit about that yeah, yeah, so, and
these are kind of the things tolook out for.
So I'd say the first thing isthere's one ethics code that has
released a statement on usingAI, and I think that we can all
use that as a good example.
So, for example I'm apsychologist APA hasn't released
(15:20):
a statement.
They're not going to release astatement for probably years,
right, cause they wait to be thelast one.
They wait for everybody else tofigure it out, and then they'll
release a very vague statementabout what we're going to do for
AI.
But it's gosh, I always butcherthe name MHCA, lmhca, mhca the
Mental Health CounselorsAssociation.
So their statement essentiallysays that AI is a new technology
(15:46):
it's okay to use, as long asyou get informed consent, and
this kind of jives with what allof our ethics codes say in
general about anything that'snew.
Right, it's okay to trysomething new, to do something
new, as long as you are checkingthat it's as secure as you can
check for and as long as you areinforming people.
(16:09):
Hey, this is a new thing.
None of us can say that we knowwhat AI will be in two years,
right, and so the key is toinform people.
I would like to use AI to writemy progress notes, and, yes, I
would recommend so.
Again, apa has not released astatement, right?
So I have no, as a psychologist, no way to say I could get in
(16:33):
trouble for using or not usingit.
But I could tell you I canextrapolate from the rest of
informed consent policies andguidance on ethics and using new
technology that I better letpeople know this is new and that
I can't know all of thepotential downsides yet.
(16:53):
So that's one thing that'sreally key is to let people know
I'd like to use this AI andthen let people know are you
using it afterwards?
Is it not recording a session?
It's just something you want touse to write your progress note
, or is it something you want touse to record a session?
Because then that can get intostate specific things where
certain states say, if you'regoing to record a session, that
(17:14):
requires its own consent as wellas different ethics codes say
that too.
So so you might have a two partthing there where you're
talking about getting consentfor AI and then also getting
consent to record them.
So think about both of thosethings and, as you're hearing me
talk about this stuff, ifyou've already played around
with AI and you're like oh mygosh, I haven't done that.
(17:36):
Don't freak out, it's okay.
Just do it the next time yousee your clients right, that's
what we always say.
And just do it better from nowon.
And that's the thing with thisnew stuff.
You know, we will experimentand we'll learn and we'll learn.
Oh, I may have done that wrong.
Okay, let me do better.
And it's not anything that youneed to really freak out about
(17:57):
or worry about.
Speaker 2 (17:59):
So that's the thing that everybody signs a BAA.
Speaker 3 (18:03):
Yes.
So number one is informedconsent.
Right, make sure you getinformed consent from your
clients.
And I would say the other partof the MHCA guidance that I
think other ethics codes willprobably agree with is that if
your client says no, I don'twant you to use AI, you can't
then say I can't see you Like.
(18:26):
Ai is not a standard of carethat you can use to excuse not
using it or getting rid of aclient for not using it.
So if a client feelsuncomfortable with it, you have
to be OK with saying, all right,I won't use it.
So the next part is technologywise.
How can you know what is safe?
All you're required to do is tohave a BAA with that
(18:49):
organization.
So that's with pretty much anytechnology.
Ai is not different in thisregard.
Think of it just like an EHR,your electronic health record,
your electronic health record.
You are giving it very personalclient data.
You're giving it client records.
You're giving it clientinsurance numbers, date of birth
, everything right, and youdon't personally research that
(19:12):
technology to see if you feellike they are doing all of their
data security the right way,because we don't have degrees in
that.
We don't know how to do thatthat's the point of a BAA, a
business associate agreement.
That that's the point of a BAA,a business associate agreement.
That is a tech company, anytechnology, says we understand
that healthcare informationneeds to be more secure.
We understand that there isthis law called HIPAA.
(19:38):
We know it exists and we arecreating our technology to
coincide with that law and to bein compliance with that law.
And so we are giving you alegal agreement, a BAA, that
says we are protecting thisinformation in compliance with
HIPAA.
So all you need, as thepractitioner, is that BAA.
That's what makes you compliant.
Now, how you get the BAA is itvaries.
(19:59):
So for some of these companiesyou have to request it.
So you might be using it andnot have a BAA, and technically
you are not HIPAA compliant.
The software is still HIPAAcompliant, it's doing the same
thing.
They're not changing thesoftware as soon as you download
a BAA right, but you are notHIPAA compliant without that BAA
.
Some of them automatically justgive it to you, send it to you
(20:23):
as soon as you sign up.
Some of them require you tosign something when you sign up
and then it's downloaded.
So it's all different.
But you do want to check andmake sure that you have a copy.
It should be a physical copy ofthat BAA, and so you typically
get that during the sign upprocess.
(20:43):
But if you don't then reach outand ask for it and if they
can't give it to you, then youare not being HIPAA compliant if
you do not have that BAA.
And I emphasize that becausethere are companies
unfortunately there was one thatis now HIPAA compliant that for
a long time was not.
They were not giving peopleBAAs and they were saying they
(21:06):
were HIPAA compliant becausetechnically they're HIPAA
compliant.
But you are not HIPAA compliantif you don't have that BAA from
them.
So don't be duped.
That's the only thing you haveto do is get that BAA.
Speaker 2 (21:21):
Is there?
We have a.
I'm skipping around in thenotes because you're answering a
lot of this or not.
Notes the chat.
Is there any?
We have a.
I'm skipping around in thenotes because you're answering a
lot of this or not.
Notes the chat.
Is there any way you couldscreen share and just show us
what a note would look like fromAI?
I mean, you said you had somesample sessions or things that
you do.
Speaker 3 (21:40):
Yeah, let me pull up.
Let me pull it up before Iscreen share.
I gotta dig around through myyou know see all these handouts
that I have.
So I do have a handout Actuallyit's somewhere else.
Let me just search for it tofind it more quickly.
But I have a handout that hasall of the different progress
(22:00):
notes I have generated fromdifferent AI progress note
generators and what I use.
The session that I use is thesession with Carl Rogers and
Gloria.
If any of you are familiar withthat, it's a very popular
(22:22):
common training session from the60s and I'll go ahead and share
my screen if that's okay.
Then If any of you are familiarwith that, it's a very popular
common training session from the60s and I'll go ahead and share
my screen if that's OK, thenhold it up, all right, so, and
I'll share this link with youguys too, so you can reference
(22:43):
it.
This is a handout that I offerfor anybody as part of that
YouTube series anyways.
So I have some important noteshere and I have comments on all
these different platforms.
There are many platforms thatare not listed here yet.
I will say AutoNotes is onethat was not HIPAA compliant
(23:09):
previously and now is.
That still gives me a littlecaution that they at one point
were not and we're not quiteupfront about that.
So they're not my first that Iwould recommend.
But here's an example of atherapy session note, of adapt
note, data, assessment and plan.
As you can see, this is areally long note.
I mean, look at this, look atthe plan section.
(23:30):
Your plan section does not haveto be that long.
Okay, I'll just like give youlike the quickest note feedback
ever and say that, unless thisperson was suicidal and we
created a whole plan for what todo and they had never been
suicidal before, right, and sothis is like a really big deal
and we're making a wholedocumentation.
You know, we're changing ourtreatment plan and do all kinds
(23:52):
of stuff.
Your plan doesn't need to bethis long.
It could be one or twosentences really.
So here I emphasize in theorange these are kind of like
caution things where it madethings up.
So here's a good example, andthis is not to pick on auto
notes, but to just say all ofthese AI platforms do this.
It says moving forward, thetherapeutic plan will focus on
(24:15):
helping Gloria navigate herfeelings of guilt, et cetera, et
cetera.
There was no therapeutic plandiscussed.
This was a first and onlysession with this client.
Therapeutic plan, treatmentplans anything like that was not
a thing.
I don't think they even had todo that in the 60s, right?
So this was not discussed inany way.
But AI this is a general rule ofAI.
It wants to make you happy.
(24:35):
It wants to fulfill yourrequest.
This is a general rule of AI.
It wants to make you happy.
It wants to fulfill yourrequest.
And if it has been trained bythe developers to say it needs
to include certain interventionsor that always mention a
therapeutic plan, things likethat, it's going to do that,
even if that's not what actuallyhappened.
So it will create these kind ofclinical sounding things that
(24:58):
didn't exist.
And I want to emphasize that,because one of the big things I
hear from people when they startusing AI is like, oh my gosh, I
could never write a note thisgood.
And a lot of you might belooking at this and like, oh, I
want to write down what it saysor how it phrased it.
You don't even know if this isaccurate.
It could be totally makingstuff up.
You don't even know if this isaccurate.
It could be totally makingstuff up.
So this is not for me.
(25:22):
This is actually not new.
This has been going on and eversince I've been training people
on writing progress notes.
It's not about making a notesound clinical.
No one is grading you on that.
We want to see what actuallyhappened in the session, and I
would rather that a 12 year oldcould read your progress note
(25:44):
and understand what happenedthan for you to overcomplicate
it by putting in a bunch ofclinical jargon.
So it is much more importantthat progress notes are just
clearly read and understood.
Let me show you one that looksreally different.
So some of them, like I said,will give you only options like
this, where it's a big field oftext, and then typically what
(26:06):
you would do is you would copyit all and paste it into your
EHR or whatever record systemyou have, and so you might tweak
it, adjust it, only copy pasteother parts over, et cetera.
Actually, I will highlight herebecause I'm going to highlight
this in the next note too whereit says Gloria is currently
facing sorry that doesn't makeit easier to read, does it
(26:29):
Significant emotional distressprimarily centered around her
feelings of guilt and fear ofrejection.
Her mood was anxious and hertearful affect suggests a high
level of emotional turmoil, etcetera.
If you have seen this session,I would not say that Gloria
exhibited significant emotionaldistress or that someone being
(26:52):
tearful suggests a high level ofemotional turmoil, right, would
we say that about someone who'scrying because their father
passed away and they got tearful?
No, that's normal, right.
And so I will say that some ofthe AI platforms because they
are trying to incorporate whatthey know we need, which is
(27:16):
often to show medical necessityit does tend to over-pathologize
or overemphasize things andit'll take what might be a
stressor or a worry or a concernand it then blows it up to be
like significant distress.
And there is a balance of yes,we do want to show things and
(27:37):
document well, for insurance,but I wouldn't want someone in
this session she primarily istalking about she lied to her
daughter about having sex, andher daughter is nine, I think,
and so I think we could.
There are different people whocould argue that, like, that is
totally appropriate to lie toyour daughter about having sex
(27:59):
when your daughter is nine yearsold, right?
And then there are people whocould argue well, it actually
could be a really healthyconversation to have.
All of those could be right,answers right.
The point is that's a reallynormal parenting experience and
she was distressed about it.
She's more distressed about,like, essentially, all the
(28:20):
things going on.
Am I going to mess up my kid?
Is it going to mess up my kidIf I tell her I lied?
I feel guilty, but is it, youknow?
Should I just deal with that onmy own?
These are normal parentingexperiences that people have,
right, um?
And so I worry that some ofthese notes will overemphasize
things like that and just kindof inaccurately reflect
(28:44):
someone's actual level ofdistress.
So I wanted to mention that.
Let's look at.
Mentalic is another one that Iwill say they win shortest note
award.
So although it looks long, youcan see it's like broken up a
(29:06):
little bit better and they don'thave giant paragraphs of things
.
Um, it is written very similarLike.
It looks more like a psychiatrynote than a psychotherapy note
to me, but it still has prettygood.
Here it said client isstruggling with adjustment to
single life after divorce, guiltover lying to her daughter,
concerns about her relationshipwith her partner.
(29:27):
There's no partner, there's noconcern about relationship with
a partner.
They just made this up.
Single life after divorce, yes,but no, that's not there.
So there are times where, again, I don't know if the AI took
her talking about her divorce tokind of mean partner, how
(29:48):
that's working.
But these companies, you haveto read these notes and you have
to edit them and unfortunately,you have to read them in a.
You have to edit them andunfortunately, you have to read
them in a fairly detailed way inorder to know if it's accurate
or not Right.
So, like this one phrase Iwould take out.
The rest of it is fine, butlet's just browse through this
(30:12):
note and then, like I said, I'llgive you guys this, this
handout, and then this handoutis the one that I update every
time I do a new review, soyou'll continue to get those
updates if you follow along withthat YouTube series.
(30:32):
Then interventions it kind ofbreaks things out into approach
and modality, actualinterventions here I'll pick on
mentalic because it's still apretty good note.
Here it's kind of doing theopposite.
The therapist used activelistening, reflection,
validation and empathicresponses to help the client
feel heard and understood.
This is Carl Rogers.
100%.
This is how Carl Rogers works.
(30:53):
This is an accurate statement.
This is not what an insurancecompany wants to see for your
therapeutic interventions, right?
So here we're constantlybalancing these two things, and
this is why I call documentationan art.
This is constantly the art thatwe're doing is balancing both
(31:14):
insurance type statements,progress, what insurance wants
to see, and what actuallyhappened.
Now I think that we can alsoidentify other interventions
that Carl Rogers presented thatare more insurance appropriate.
Here I'll give you a littlesidebar tip.
So how do you know if yourinterventions are appropriate
(31:39):
for insurance or not?
I like to use what I call thegrandma rule.
So if you, assuming you have agood relationship with a you
know a nice grandmother, withyour nice grandmother, how would
you expect her to listen to you, or how do you expect
interactions with her to go?
Are you expecting her toactively listen?
(31:59):
Probably.
Are you expecting her to beempathic?
Probably.
You can also extrapolate thisto your best friend.
How would you talk to your bestfriend about something?
Are they validating?
Yes, you don't need a graduatedegree to do any of those things
, right?
Do we hone those skills ingraduate school?
(32:20):
Absolutely.
Are those necessary foreffective therapy?
Absolutely, but those arebaseline things that you don't
need a graduate degree to do.
So for insurance.
That's why I call it thegrandma rule.
What is something that you hadto go to graduate school to
learn to do?
Those are the types ofinterventions that you want to
(32:42):
focus on having in your progressnotes.
So here, mentalic kind ofover-pathologized the client and
then kind of underrepresentedwhat the therapist did from an
intervention response.
But I will say, with a lot ofthese things you can teach these
platforms what you want it tosay.
(33:03):
So there is one Clinical Notes,ai, where you can give it a lot
more direction and you can tellit, for example, choose five of
these interventions to put inmy progress note and you could
give it a list of 50interventions and that way you
know that at least it's fromkind of an approved list that
you've curated.
(33:23):
All right, let me stop sharingand just check in here and see,
see how everybody's doing andwhat you guys think.
Speaker 2 (33:31):
Well, one of the questions I saw.
But so in, I think this is.
But so and I think this isdovetailing on what you're
saying it's not really able.
Like with ChatGPT, we could putin something we had written and
ChatGPT will start giving usresponses that sound like us.
So this isn't like that.
We're not able to train it thatwell yet.
Speaker 3 (33:54):
It depends on the platform.
So this is why I need tore-review.
So, for example, clinical NotesAI.
I think you can do that alittle bit more.
Session Aid is another one, butthe way you're going to do that
really well is by using aplatform where you can give it
(34:14):
those really specific directivesfrom the beginning and then
maybe adjusting your templatebased on how it's writing.
I do think Orchid is one that'san EHR and an AI platform and
(34:38):
that one does learn how you workover time because it has a lot
more information from you,because it's literally
integrated within the EHR, whichis likely the future of how
this will work, and so I dothink that is what this will
look like in two to five years.
But right now it's a lot moreeffort and just like if you
found with ChatGPT, you stillhave to do a lot of work to oh,
I forgot to tell it this, oh, itgave me this and I realized I
(35:00):
actually want somethingdifferent, and then you have to
rework it.
So you're kind of doing thatwith your template and with the
directives you're giving the AI.
Speaker 2 (35:12):
Okay, One question was we hear a lot of scary
information about AI taking theplace of clinicians in the long
term.
Speaker 3 (35:19):
What are your thoughts on that?
All you have to do is readthese notes to know that's not
happening anytime soon.
It's not.
I mean, I train people on howto write progress notes.
I'm not worried about AI takingmy job, so it's definitely not
taking your job.
I do think, and why I'm excitedfor this next round of the
(35:41):
reviews I'm doing in the nextmonth.
A lot of the companies aregetting on board with kind of
this integrated approach of theclinician.
Can, you know, use the AI towrite progress notes and send
messages to the client, sendhomework assignments to the
client, and the client can login too and check up on things.
(36:03):
So I know that young sidekickis one that I'm about to test
that offers that already.
Ooh, and I'm going to blank onthe other one.
So I feel bad, but they'll allget YouTube videos so you'll be
able to see those.
But I'm really excited aboutthat because I think that's the
reality of how we can use theseand how clients will be able to
(36:24):
use them.
So think about, in one of thegroups that I run, people were
talking about some people weretalking about the benefits of
doing sessions every other week,for example, and one therapist
I work with was saying they workwith a lot of trauma and a lot
of times, sessions are reallydeep and intense, and so they
actually find it more beneficialto take a break and do sessions
(36:47):
every other week.
Well, if you could add on an AIcomponent, where then the AI
knows to like follow up with aclient, maybe a week later, or
give an assignment, and thenthey can log in and have some
kind of an app that they canreference To me, that would be
hugely beneficial to thetherapeutic process and I think
(37:08):
a lot of people would like that.
So I'm excited for that.
And where it's an integrated,clinician can use it for these
things and client can use it forthese things, and client can
use it for these things.
Related to that, I do want tobring up one other caveat that I
think is really important andhasn't been brought up as much,
which is, as I'm talking aboutthis, think of all this data
(37:31):
that's going back and forth inthis AI platform and this is the
AI platform.
In this AI platform and this isthe AI platform.
It's not your EHR and it hasmaybe it has a full transcript
of every session.
It has maybe client reportsthat they're submitting to you.
It has now also progress notes,maybe three versions of the
(37:52):
same progress note that yourevised or that you had it
produced for you in differenttemplates that you revised or
that you had it produced for youin different templates, and so
now you essentially have clientPHI in two systems and you're
maintaining it in two systemsand they could be two HIPAA
compliant systems.
That's great.
(38:12):
What happens when that clientsubmits a records request and
they say I'd like a copy of myrecords and you give them a copy
of the records and they saywell, what about the transcripts
?
Isn't it giving you atranscript?
Isn't it producing a transcriptof the whole session?
I worry that that is going tohappen one day to an
(38:35):
unsuspecting therapist who willthen realize oh like, releasing
a whole transcript to someone isa little different than
releasing a progress note andit's their record.
And in the US and in Canada aswell, it's really difficult to
(38:56):
tell someone you can't accessyour own record.
So think about if you went to aphysician and there was a
transcript of that session andyou knew there was a transcript
and you said I'd like a copy ofthat and the physician said no.
That now becomes a client rightsissue, right?
I think that the law in mostcases and in most states would
(39:18):
side with.
People get access to theirmedical records and personally I
think that's probably the rightcall.
Right, people need access,should have access, to their
medical records.
So what does that mean?
For us, that means, if you wantto use one of these platforms,
that that is a question to askthem.
When you sign up, do youautomatically delete these
(39:41):
things?
So some of the platforms willlet you create that as a setting
.
So at once a week it's deletingall the notes or it's deleting
all the records.
Some of them will let youoverride that.
Some of them automaticallydelete that.
So session aid, I think, is onethat is very transparent about.
We do not plan to be an EHR.
We do not plan to maintain thisinformation long term.
(40:04):
It's going to be regularlydeleted.
That could be really beneficialfor managing this issue.
It could also be a realstruggle if you're using it
because you tend to get behindin your progress notes and you
want to have access to thoseprogress notes two weeks later,
right?
So that's why I'm bringing itup, as it's not necessarily a
(40:24):
good or bad thing, but you doneed to consider that ahead of
time and what you're going to dowith it and what the platform
offers you in the way of optionsfor deleting or saving the
information options for deletingor saving the information.
Speaker 2 (40:44):
So one of the questions here would love to
just know how we can bestutilize AI in an ethical way,
then mostly to save time withtreatment plans, you know.
Is there like a hybrid way touse it, Like if we're not just
recording everything, then howwould you use it?
Or how would you, how do yourecommend that we use it?
Speaker 3 (40:59):
Yeah, I think this is where I don't think it's always
the solution and it reallydepends on what your particular
problem with progress notes is.
If it's just time, it's rarelyjust time, I will say.
Usually the time issue isactually it takes me 20 minutes
to write a progress note becauseI sit there and question
everything I'm writing andrewrite it.
(41:20):
So that's time is not yourissue.
It's feeling confident in whatyou're writing.
That's actually the problem tosort out there, right, and I
would say that's the most commonscenario I see.
And if time is your issue, isit that you're ending?
Are you ending sessions on time?
That's not going to change withAI, right?
(41:40):
Are you giving yourself enoughtime to write progress notes?
Do you have time in yourschedule for progress notes?
You're still going to need thatwith the AI platform because
you'll have to review that,review them and then copy paste
it into your EHR, right?
So, really, you know, I dothink it's the first thing I
always recommend for everyone islook at what is the real reason
(42:02):
that I'm struggling withprogress notes and try to
address that first.
If it is genuinely time, whichI do understand, that is a big
part of it for many people.
That is where this hybridmethod can be really helpful.
That is a big part of it formany people.
That is where this hybridmethod can be really helpful.
But for me, the hybrid method,I think, is more effective if
(42:23):
you are literally thinking of itlike collaborative
documentation.
So for treatment planning, forexample, I always recommend you
never need to do a treatmentplan outside of a session.
The treatment plan is literallywhat you and the client are
agreeing to do in therapy.
There is nothing more valuablethat you could talk about with
(42:44):
them, right, you want to?
Literally.
It's an opportunity toliterally be on the same page
about what you can accomplishtogether, what they can expect
from you, what you can expectfrom them, where you want to be,
where you want to work andfocus on.
I don't think that you need tosit there and you know, like
(43:05):
some of us have done in agencydays, and spend half an hour
filling out a really long formwith them, but absolutely
talking to them about the goalsthat you want to work on and
identify like why are you here?
What is it that you would liketo work on?
There is no question, based onresearch, that that is a
(43:29):
valuable clinical discussion tohave, and then having that
clinical discussion will have apositive impact on the therapy
you do with clients and it'sempowering for people.
So to me, that is something youdon't really need AI to help
you with.
Now, where that might come up isyou say, okay, but how am I
supposed to sit here and writethis with a client when I'm
thinking of what to write andI'm trying to produce this
(43:51):
content and, yes, you're saying,make it understandable, but I
still got to write anintervention, right, and it
takes me some time to think ofthat.
And so that's where I thinkthat having cheat sheets and
templates and I think that wasone of the questions that came
in ahead of time too, that youhad sent me that having like
(44:12):
pre-created statements is themost valuable thing.
And that way you're just eitherfilling in the blank or you're
circling things or you'rechecking off boxes, but they're
detailed, they're not.
When I say checking off boxes,I don't mean, like what your EHR
gives you as the standardcheckbox interventions list.
That says cognitive behavioraltherapy that's a thousand
(44:34):
interventions, right.
We're not checking that off asan intervention.
Cognitive restructuring that's amore specific intervention.
Creating a pros and cons listintervention, creating a pros
and cons list.
That's a more specificintervention.
(44:56):
You know, reframing, groundingwork, deep breathing exercises,
exploring patterns in familybehaviors, exploring patterns in
the family.
I would say Carl Rogers didthat in that session with Gloria
, but that's not one of theinterventions that any of these
AIs are picking up from thatsession.
But he absolutely did that andtalked about how her
relationship with her father ishaving an impact on this
(45:19):
situation she's having with herdaughter, and so it was a very
active reflection.
So, to me, curating a list ofstatements that you use and of
phrases that you use that applyto the way you work and the
clients you work with is themost useful thing.
And if you can use that to thengive one of these AI platforms
(45:43):
a template and a cheat sheet,then it's already going to sound
like you and that's kind ofbypassing what you're talking
about with, like using chat, gptand having it teach you how to
write.
So using one of the platformswhere you can give it a lot of
direction and you can reallycurate the template to your own,
I think is the best bet.
Speaker 2 (46:08):
Well, and I think that helps with the next
question as well, because whenyou talked earlier about the
quality of the note and it, youknow you talked about, you know
making up things.
Well, if we're trying to alsocreate a note that protects our
client, right, because of thethings like we're in Texas and
there are things that ourclients are going to discuss and
we don't want to put that inthe note, and so I'm hearing you
(46:33):
say that a template with thesesort of responses that you
usually use could really behelpful.
Speaker 3 (46:40):
Yeah, and I always recommend that too, when I've
done a couple of YouTube livesas well on this particular topic
, like, how do you documentsensitive topics and this has
become much more of an issueover the last few years right,
and so having I think, whetheror not you use AI having some of
(47:01):
these statements ready to go sothat you don't have to think
about them in the moment whenit's stressful, is really
helpful.
So I'll give you some reallyspecific examples.
So, let's say, your client istalking to you about their
feelings around having anabortion and, regardless of the
state you live in, that could bea really heavy and important
(47:25):
topic to your client and is areally emotional experience,
regardless, right or could be,and so that's pregnancy loss,
and pregnancy loss allows you tocapture the emotion around the
event without actually talkingabout the event itself and how
it happened, actually talkingabout the event itself and how
(47:49):
it happened.
But having some kind of likepre-created statements like that
for these highly sensitivetopics will help, and I
encourage people you know youcan watch one of my YouTube
lives to get information on thatto get ideas for what to write.
Talk with your colleagues aboutwhat they write, but the fact
is, if you have a client who'sreally stressed about that
experience, it might besomething they're talking about
for months, right?
(48:09):
So it's not a matter of like.
Let me be as vague as possibleand never document this.
That might be an extremely thatis an extremely relevant thing
to document for many clients,and so having an idea of how to
phrase it is really, reallyimportant.
Speaker 1 (48:28):
Excellent.
Speaker 2 (48:31):
So, if it's okay with you folks, if you have the hand
, raise emoji or if you want topost in the chat, I'd love to
open it up to your questions Ifyou'd like more detail about
things that Dr McCaffrey'stalked about or just more
information.
So Alyssa says how to phrase itverbally as it is recording or
(48:55):
in the edits.
Now, alyssa, if you unmute,you're going to be on the
recording.
Speaker 1 (49:01):
So I totally get it if you don't want to do that,
but I'm fine with that.
Okay, I don't care.
My concern is so.
I'm just like I hear all theHIPAA and I hear the business
associates, agreements and allthese kinds of things, but in
the end of the day, someone islistening into my session,
whether it's AI or not, and thenthat's being transcribed.
So when you said phrasing, areyou referring to as we're
talking to them to change ourdialogue?
(49:21):
Are you referring to when weget the note that we have to
edit and everything that we'rehaving go-to statements for that
?
I was just unsure about thatpart.
Speaker 3 (49:29):
Yeah, really good question.
So that might be aconsideration for whether or not
you use recording, Because tofollow up with that, you know
very real example of a clientdiscussing concerns around an
abortion, right?
Maybe they're deciding what todo.
Whatever that might be, Oncethat word pops in, it becomes a
(49:49):
very charged issue.
So is your AI platform going topick up that word in a
transcript?
Probably and that may or maynot be something you can that's
likely not something you cancontrol, so that's something
something to think about.
Now, that's absolutely.
You can delete the transcriptimmediately or make sure that
(50:13):
you're using a platform whereyou can, right, Because maybe
your practice is just you deletethe transcript immediately for
all sessions, not a big deal.
But the AI needs some data togive you a progress note, right,
and the more data it has, thebetter note it will give you.
Just like with anything, Icould give you a better progress
(50:34):
note if I have more data aboutthe session too, right?
So anything that's, it's alwaysgoing to work that way.
So if you, you could either notdo the recording and you could
curate the information you giveit, and that way you can just
straight up use the wordpregnancy loss from the
beginning.
It never hears anything else,and that's what the AI will use.
Or you can wait for it to giveyou the information and then you
(50:57):
edit it in the note, but thenyou would want to make sure that
you delete any transcript thatthere is.
And I will say you know, andhaving a transcript isn't always
all bad.
If you do supervision, thatcould be awesome, because you
get way better information aboutwhat's happening in sessions
(51:17):
than you will secondhand fromtalking to trainees, right?
So you know, it's not that it'sright or wrong, but it's
important to think through allthe potential ramifications
Awesome, Thank you.
Speaker 2 (51:33):
Yeah, Keisha wants to know is there one platform that
you found to be the mostaccurate in capturing the
session without making things up?
And you said I know, you saidyou're re-reviewing these things
, but what have you found?
Speaker 3 (51:46):
Yeah, no, I haven't really.
They all like they all madesome kind of mistake at some
point.
I forget which one.
Like one of them, even this isa good example of how AI is not
going to take your job.
Okay, so one of them literallyphrased something how Carl
Rogers did.
(52:08):
But guess what Carl Rogers did?
He reflected back to Gloriawhat she had said and she had
said I forget exactly the phraseit was, but I would find it if
I could remember which platformit was but she specifically said
something like you know, mydaughter's going to hate me if I
do this.
And he said you're afraid yourdaughter's going to hate you if
you do this.
Right, he just reflected itback to her, which is what a lot
(52:32):
of us would do.
And so then the the, the noteit generated was like you know,
therapist told client that herdaughter would hate her if she
did this.
So it took it out of context,right, so in context, like I
never would have even thought tophrase it that way.
But then the AI, it's just anAI, it doesn't know.
So things like that are anexample of how it didn't get it
(52:52):
wrong.
It wasn't necessarily wrong,but it was out of context, and
so those are the things that youhave to look out for.
Speaker 2 (53:01):
There was a question earlier about shredding a note
once you put it into AI, and Iknow we've talked about deleting
notes and things like that.
I want to be really clear.
There is nothing in the LPC,lmft, bhec, social work.
Well, actually I take that back.
Lmft speaks to how we get ridof our notes, or at least it did
(53:22):
.
I don't know if they do anymore.
So this is all best practicesat this point, right?
Until we have more precedent,until we have things written
into the rules.
So this is the best we've gotright now, right, right, melissa
?
I mean.
Speaker 3 (53:38):
Yeah, and I would also say I think it can be
helpful because a lot of thisstuff is so vague and this is
literally why I have a job,because documentation
requirements are vague so it canbe really helpful to make it
less vague by like what is yourprogress note?
And really define that.
What is your progress note?
If you use an EHR, it'stypically whatever note is
submitted to that EHR.
(53:59):
So if that's my progress note,any other notes I have floating
around need to be gone.
So if that's a transcriptsomewhere, if that's the
progress note, that's in the AIplatform.
That is unedited, because then Iput it over here and edited it
and maybe I edited it a littlebit over there, but then when I
copy pasted it, oh, I didn'trealize that and so I made a
(54:20):
change over here.
So now I have two differentversions of the same note.
Right, I mean, that wouldabsolutely happen.
It could get very confusing.
The easiest way is just once Ihave that progress note in that
system, I get rid of all myother notes, and so you could
also shred.
Yeah, if you took notes, if youtake tons of people take notes
by hand, totally fine, you canabsolutely do that.
Take your notes by hand andthen, as soon as it's in an EHR,
(54:43):
shred that.
Or in your computer somewhereelse, shred it.
So kind of make a rule foryourself about where you're
going to have your notes.
What is the note, and just keepthat one, okay.
Speaker 2 (54:57):
We're getting requests here if you have sample
informed consent.
So this is a good time, I think, for you to drop some links
into the chat and everybody.
You know the drill when you seea link, click it, Because once
the meeting ends, the links goaway and then you'll have that
tab open amongst all thehundreds of other tabs I'm sure
you have open.
What's the best way to contactyou?
Speaker 3 (55:21):
Yeah, so the best way , I will put in the link for my
website and then I will will putin the link for my website, um,
and then I will also put in, um, the direct link to the book as
well.
So, um, my business called QAprep.
Here's the website, qaprepcomfor those listening later, so
(55:42):
just qaprepcom.
Uh, takes you there.
The book is called Stress-FreeDocumentation for Mental Health
Therapists.
There is a paperback versionand a Kindle version and Amazon.
It's available on Amazon.
Amazon randomly puts things onsale and it is on sale right now
(56:06):
and I still get the same amountof money.
So go buy it on sale.
And it is on sale right now andI still get the same amount of
money.
So go buy it on sale.
Um, yeah, um, and that book has.
I'm going to put the name ofthe book in the chat stress-free
documentation for mental healththerapists um you've got a
follow-up.
Speaker 2 (56:24):
Has been following your your youtube chin page oh
exciting.
Speaker 3 (56:29):
Thank you, keisha.
Yeah, and in in the book it hasall of my templates, it has my
full paperwork packet, and sothere's also links in the book
so that you can log in and getthem electronically.
So you don't have to sit thereand photocopy or copy copy into
your own.
You can just download the formsand and use them directly.
(56:49):
So that'd be the book I wouldsay is the easiest way to get a
copy of of my all my consentforms.
Speaker 2 (56:56):
Excellent.
I love that, thank you, thankyou.
Thank you so much, andeverybody.
This has been wonderful.
I'm so glad you're here.
Step it uppers.
If you are in the membership,you will get this as an
asynchronous CE.
Once we process it, y'all knowwhat to do.
Go visit Dr McCaffrey, get yournotes in shape, get rid of the
(57:17):
get rid of the shame of gettingbehind that I know nothing about
.
And, yeah, enjoy your Mondayevening.
I think you made us all feelbetter.
Dr McCaffrey, thank you so much.
Speaker 3 (57:28):
Good, good, I'm glad.
Thank you all for yourquestions.
That helps so, so much and Ireally I value your trust and
vulnerability there.
So thank you all.
Speaker 2 (57:38):
All right, we'll see you later.
Guys, have a wonderful evening.
Bye.
Popular Podcasts
24/7 News: The Latest
The latest news in 4 minutes updated every hour, every day.
Stuff You Should Know
If you've ever wanted to know about champagne, satanism, the Stonewall Uprising, chaos theory, LSD, El Nino, true crime and Rosa Parks, then look no further. Josh and Chuck have you covered.
Dateline NBC
Current and classic episodes, featuring compelling true-crime mysteries, powerful documentaries and in-depth investigations. Follow now to get the latest episodes of Dateline NBC completely free, or subscribe to Dateline Premium for ad-free listening and exclusive bonus content: DatelinePremium.com