29. Using AI for Psychotherapy Progress Notes: Pros, Cons, and Ethical Considerations

Using AI for Psychotherapy Progress Notes: Pros, Cons, and Ethical Considerations === Meghan Beier, PhD (2): [00:00:00] Welcome to today's episode where we are going to explore the use of artificial intelligence in mental health note writing As a therapist, social worker, or psychologist, we know that accurate and thoughtful documentation is a vital for a variety of reasons, including tracking patient progress, for insurance reimbursement, and for legal purposes. Meghan Beier, PhD (2): So we need to ensure that our documentation and chart notes are meeting legal and ethical standards. However, the administrative burden of notetaking, especially for so many seemingly conflicting audiences, can be really overwhelming. Meghan Beier, PhD (2): This is the appeal, at least for me, of AI note writing tools, which claim to streamline this process. But the question becomes, should we adopt these tools? There are some potential benefits, but we need to proceed with caution and thoughtfulness. If we are going to adopt them, how do we ethically integrate these tools into our [00:01:00] practice? Meghan Beier, PhD (2): I'm gonna go through a bit of background about AI. The landscape of AI tools and mental health charting as it stands right now. And then I'm going to introduce an expert, our guest, Dr. Melissa McCaffrey, who is an expert on ethical documentation for mental health professionals. ​ What is AI? --- Meghan Beier, PhD: Let's start with defining what is AI? AI or artificial intelligence refers to computer systems that can perform tasks, typically requiring human intelligence. These tasks include things like recognizing patterns, understanding language, making predictions, and even generating text. In the case of note writing. Meghan Beier, PhD: AI uses something called Natural Language Processing, or N-L-P. NLP allows a computer to understand and process human speech and text [00:02:00] when you input spoken words or written notes. AI analyzes the language, identifies key points and generates text based on patterns. It has learned from vast amounts of data. Meghan Beier, PhD: Some AI models use machine learning, meaning they improve over time by recognizing the way users structure their notes, and then they refine their suggestions accordingly. Others rely on predefined templates and algorithms to create summaries based on the information provided. Meghan Beier, PhD: However, AI is not perfect. It doesn't think the way humans do. Instead, it follows patterns and predictions based on data. This is why we need to be thoughtful about its use, and if we decide to bring it into our practice. There needs to be human oversight to ensure the accuracy and appropriateness of AI generated notes. Common Features Found in AI Note Writing Tools --- Meghan Beier, PhD: Now I'm gonna go into some of the key features that are found in many, if not [00:03:00] all, of the note writing tools currently on the market. I wanna highlight here that I am not being paid by any of these companies. I have no financial stake in any of them, and I'm not getting any discounts.So there's a few different features. One is automation of progress notes. An AI note writing tool can generate or write a comprehensive session note by analyzing audio recordings or transcripts. Sometimes this analyzation occurs while the AI note writing note writing tool is listening in on a live session. There can also be integration capabilities, so some AI note writing tools are being built into existing electronic health record systems. So as an example, there is, a EHR system called EPIC, which is used in many large healthcare settings. According to a Washington Post [00:04:00] article, EPIC AI tools are being used to transcribe approximately 2.35 million patient visits per month and draft 175,000 patient messages each month. Meghan Beier, PhD: So this is already being integrated into some of the tools that are currently on the market in terms of EHRs. many of these tools also offer customization. So, for example, many AI tools, can allow you to customize the templates to fit different therapeutic approaches or personal documentation styles. Meghan Beier, PhD: For example, you could choose between a soap note or DAP note, or maybe even just provide a simple summary of the notes that you provided. and many of them advertise that they follow privacy regulations. So they, talk about ensuring patient and client confidentiality. [00:05:00] and they are designed to be HIPAA compliant. Meghan Beier, PhD: But please listen in to my conversation, with Dr. McCaffrey because she talks about this a little bit more in the sense that what we as mental health providers think of as HIPAA compliant and following privacy regulations may not be exactly the same as what some technology companies perceive as privacy. Popular Note Writing Tools for Psychologists --- Meghan Beier, PhD: So now I'm gonna talk a little bit about some of the AI note writing tools that are currently available. While I'm not gonna go through every single note writing tool on the market, because there are a lot of them, I wanna provide a few examples to highlight the differences you may find among tools. Meghan Beier, PhD: Again, I just want to highlight that I'm not being paid to advertise any of these tools. This is simply to provide a few examples. Additionally, the space is changing rapidly and tools are changing or adding features quickly. So what I share here today may not be the same in six months or a year. [00:06:00] So one of the tools I wanted to start with is called AutoNotes. Meghan Beier, PhD: this used to be, and this is an example of how things are changing rapidly. This used to be just a, a preset of checked boxes. So after a session, a therapist could go through and click, different boxes that were relevant to the session, and a note would be generated from the option options chosen. Meghan Beier, PhD: However, they've recently added features like being able to record sessions, dictate a summary, or provide written text. However, the checkbox option may feel more comfortable for some therapists that are looking to provide more general notes without a lot of specific details about their patient. Quill Therapy Notes is a tool that was started by a husband and wife team, Jen and John Sester. Meghan Beier, PhD: sorry if I didn't say that name, last name right. Jen is a licensed therapist with her own private practice. John is a computer engineer. They partnered to develop the tool, and what's unique about this one is that it [00:07:00] requires the therapist to provide a dictated or written summary of the session. It does not record or listen in on any therapy sessions, like some of the other tools that I'll mention next. The tool also does not store any data, so a note is generated from the summary that you provide. and then that summary or the note that's generated is needed to be copied and pasted into your EHR. The dictated information and your written summary and the note generated are all erased. Meghan Beier, PhD: This is really good for privacy, but you have to be careful to get through the full process with each note, or you'll have to start from scratch. Many of the other products on the market, such as Mentalyc or Supanote, provide the features described previously. For example, being able to dictate a note or provide a written summary, but they will also either record or listen in on a live session or ask for an audio file from a recorded session to analyze.[00:08:00] Meghan Beier, PhD: These may provide deeper insights, but some therapists feel uncomfortable with recording sessions or having a third party tool listen in, even with the promise of HIPAA compliance. Finally, there's some other tools that go even further. Some examples or things like Upheal and Blueprint. they go beyond note writing and they try to provide additional support to therapists by providing insights and patterns across sessions, or even providing in session guidance as the tool is listening in. Meghan Beier, PhD: Now as we are moving into the interview with Dr. McCaffrey, I wanted to first highlight a few of the ethical considerations that she and I will be discussing in greater detail. I wanna go back to the APA psychology ethics, that we all learned in graduate school. Meghan Beier, PhD: So first, Beneficence and Nonmaleficence. Meghan Beier, PhD: Psychologists should strive to protect the rights and welfare of those with whom they work professionally. So we wanna make sure that these [00:09:00] AI tools are protecting the rights and welfare of those that we are working with. Meghan Beier, PhD: Fidelity and responsibility. So psychologists have a moral responsibility to help ensure that others working in their professional. Profession also uphold, high ethical standards. So, we wanna see are these tools actually up to our high ethical standards? Are they meeting that bar that we would meet for ourselves in our own note writing? Meghan Beier, PhD: principle C is integrity. In research and clinical practice. So here mostly we're talking about clinical practice. Meghan Beier, PhD: Psychologists should never attempt to deceive or misrepresent. And in this, I kind of think that, we don't wanna deceive our patients by not telling them that we're using these tools. So we want to make sure that we are, disclosing this information either verbally. In writing or ideally both. We also wanna make sure that the notes that are coming from these AI tools are not deceiving or [00:10:00] misrepresenting what's actually happening in the course of the session. Meghan Beier, PhD: And we know that as, Dr. McCaffrey will share a little bit more, that sometimes these tools tend to put things in that weren't there. So we need to be really careful about that. principle D is justice. We have a responsibility to p fair and impartial. And so we wanna make sure that the notes that are coming from these AI tools are fair and impartial and, that we're looking for biases that might be there, that are coming from the tools themselves or some of the data that was inputted to them prior to us using them. Meghan Beier, PhD: and finally, E is principle. Principle E is respect for people's rights and dignity. so we should really respect the right and dignity and privacy and confidentiality of those that we are working with professionally. So these tools that we're using, [00:11:00] are they actually protecting the rights the dignity, the privacy and confidentiality of the patients that we're working with. Meghan Beier, PhD: There are some tools out there that say that they're HIPAA compliant and, we're finding out actually are not. So we really wanna pay attention, to how the data that is being collected by these tools, is being stored, processed, or protected. And those might be questions that you wanna ask of the company before using them. Meghan Beier, PhD: So that's just a brief overview and highlight. Meghan Beier, PhD: now I'm excited to bring us into this interview where we can dive into this in a lot more detail. Interview with Dr. Maelisa McCaffrey --- Meghan Beier, PhD: I'm excited to introduce Dr. Maelisa McCaffrey, a clinical psychologist with expertise in helping therapists with their documentation. She joins us today to talk about the ethics of using AI for note writing. Dr. McCaffrey, thank you so much for being here. Dr. Maelisa McCaffrey: Thanks for having me. I'm excited to talk about all this [00:12:00] stuff. Meghan Beier, PhD: Dr. McCaffrey, you specialize in helping mental health professionals with documentation. Can you tell us a little bit more about your work and your expertise? Dr. Maelisa McCaffrey: Absolutely. Dr. Maelisa McCaffrey: So I help therapists with all things documentation, so whether that is training, consultation on little things like how to create goals for treatment plans, how to create a progress note template, how to update policies and procedures and intake paperwork, any and all of the above. I work with therapists through giving courses, doing trainings through different associations. Dr. Maelisa McCaffrey: I have a YouTube channel where I do free training. And, a subspecialty is that I help therapists catch up on their paperwork. So I run groups called the Paperwork Catch Up Group, and we all work together and work to get notes done, whether someone is a couple of weeks behind, or many months, or even years behind. Meghan Beier, PhD: Hmm. That's excellent. I actually know a couple of friends who maybe that could be helpful [00:13:00] for so I'll have to refer them to you. Dr. Maelisa McCaffrey: Yeah. Meghan Beier, PhD: Yeah, absolutely. The reason I invited you here today is because I really wanted to talk about the pros and cons and ethics of using artificial intelligence for progress notes. Meghan Beier, PhD: so before diving into AI, What are some common mistakes or ethical issues you see in general mental health progress notes? Dr. Maelisa McCaffrey: Yeah, I mean, the first one is not doing them. So that's why we have the paperwork catch up group. and that literally is something, you know, I, I didn't start a business, because I do this full time, all of the documentation training. Dr. Maelisa McCaffrey: So I don't have a private practice anymore. I don't work at another place. I did not start this business thinking that half of my business and work would be helping therapists catch up on notes, right? But it's something that as I started working with people, I found was the most common struggle that they [00:14:00] were managing and that the vast majority of people are managing it as a deep, dark secret about which they feel a great deal of shame. Dr. Maelisa McCaffrey: And so I was doing all of these individual consultations, meeting with people, and, and we were, creating strategies who are creating a catch up plan. We were talking about, another related common struggle is, the reason that people aren't doing notes is, is needing to figure out time and their schedule that works for them to get the notes done. Dr. Maelisa McCaffrey: And so, through doing all of these individual consultations, I found, you know, it is often not just a time management issue. So sometimes, There is that, where someone hasn't really thought through, like, how long does it take me to do a progress note? So then, do I need to schedule time in my schedule, in my calendar, literally have notes time? Dr. Maelisa McCaffrey: and, and a lot of people are just assuming they'll get their notes done [00:15:00] in between sessions. And, that is not the reality for most people through my own, you know, anecdotal experience of, of working with hundreds of therapists. I agree. Yeah. The vast majority of people, it's like, you need time to decompress. Dr. Maelisa McCaffrey: You need time to answer emails or the phone or, you know, reply to a text from your kid or your mom or whomever. So, you know, all the things, go to the bathroom, have a snack. So. many people are expecting they will get the notes done then but not actually doing it and then it, it creates this whole host of other things. Dr. Maelisa McCaffrey: So, so those are a couple of the, the common mistakes that have nothing to do with the content of the progress notes, right? but I would say those are very, very common struggles that, that many therapists have secretly. And hold secretly because they don't recognize how common they are and they can be feel very embarrassing. Dr. Maelisa McCaffrey: So please, if you're listening to this, know you are not the [00:16:00] only one. You're not alone. There are lots and lots of wonderful therapists who are also dealing with the same thing. Dr. Maelisa McCaffrey: I would say the next most common, like problem or struggle that I see with progress notes is figuring out what does insurance want. Dr. Maelisa McCaffrey: So even sometimes for therapists who don't contract with insurance companies, but are offering clients a super bill, you know, and they're, they're concerned, like, I, I don't, I don't have to worry about insurance, but I am worried what would happen if an insurance company requested my notes because I'm not thinking about it, and I'm not contracted what would happen? As well as, obviously, people who are contracted are often terrified of having an audit. and, and what does insurance want to see in notes? And so, this, this fear and, and mystery around that topic, you know, leads to things like avoiding notes, not getting [00:17:00] them done, again. Or often leads to things like over, overwriting, rewriting, doubting and questioning and spending, you know, 30 minutes on a progress note. Dr. Maelisa McCaffrey: And so, that is just not a sustainable practice and it's also very stressful. So, and I, I think that kind of ties in with a lot of the AI stuff that we're going to talk about today because so many people are worried that they're not putting the right things in their notes. And I would say insurance is one of the big reasons for that. Dr. Maelisa McCaffrey: But even if people aren't worried about insurance, they are still wondering, am I putting the right thing in my notes? And it's not like you're posting your note, your progress notes on, you know, a Facebook group for people to read because it's all private, confidential information. So that lends itself to be just continuing to be more and more mysterious. Dr. Maelisa McCaffrey: And, and so people spend a [00:18:00] lot of time either writing and rewriting progress notes or worrying about them, or stressing about them, or even, like, watching my trainings and watching, like, Everyone else's trainings because I'm not the only documentation expert, right? And someone's going to hear this and say, Oh my gosh, who's the other people I need to hear what they have to say too. Dr. Maelisa McCaffrey: And so then people find all these resources and are constantly spending time updating their templates and changing what they're doing and saying, well, this person says this, but this person says this other thing. and so all of these lead to just an increased and a huge amount of time and effort and stress around progress notes that does not result in feeling more confident in what they're actually putting in the notes. Meghan Beier, PhD: Yeah, I can relate to a lot of that, especially the, the idea of what insurance needs, because I've had situations [00:19:00] where I've put stuff in my note and, I worked in a hospital system and we were doing, you know, health and behavior codes versus, you know, psychotherapy codes and, and then we got, I got dinged for one of my patients and said, this is psychotherapy, it's not health and behavior and, actually wasn't able to continue following that person because their insurance wouldn't continue paying. Meghan Beier, PhD: So, I can understand, the concern and wanting to really make sure that your notes are effective for insurance. I had a follow up to what you were saying about people being behind in their notes. Meghan Beier, PhD: you ever had consultations with owners of practices where people in their practice had gotten behind in notes? And, you know, how did you work with the owners? Dr. Maelisa McCaffrey: Yeah, absolutely. And a lot of times that is, you know, a group practice owner. I would say the common scenario for that particular person is. [00:20:00] They find out that someone is behind in notes, and they find out like three months seems to be kind of the common number where, because that is enough where it's a significant amount of time for someone to catch up, and the group practice owner is like, Oh, no, right like we need to do something now and they want to create a plan. Dr. Maelisa McCaffrey: And so, either they, they worry and schedule a consultation with me at that point, or they try to enact this plan and it doesn't work well, and then they, they schedule an appointment with me after that. and so, it is a really common issue, and I tell anyone who's a group practice owner or a supervisor, it is not, this, this issue is so common, it is not a matter of if, it's a matter of when. Meghan Beier, PhD: You have Dr. Maelisa McCaffrey: someone who will have this issue. So just know if you were in that position, please, please have some strategies in mind and expect that this will be a thing that will come up. so for group practice owners, I talk with them about, you know, having [00:21:00] policies and procedures, which is kind of obvious, but, about having those be very regular and meaning, like, have your policy be 48 hours that notes need to be due and then check up on that every single week. Dr. Maelisa McCaffrey: And often what, what happens is that the group practice owner, is either not delegating that task well, or maybe has too much on their plate. And doesn't have the support they need to actually run those reports. Cause that's a pretty simple report. Most, most practice owners can run to find out, okay, how many notes does everybody have due? Dr. Maelisa McCaffrey: and so if you catch that issue on a Monday for last week, it's not as big a deal as if you catch it three months later. Right. Meghan Beier, PhD: And Dr. Maelisa McCaffrey: so I say like the common issues for group practice owners is one, just assuming that everybody is doing their notes. And you can never ever assume that and never assume that everybody knows what they're [00:22:00] doing. Dr. Maelisa McCaffrey: It has nothing to do with the level of experience. People that I work with are all the way from recent graduates out of grad school to have been licensed for decades. And are literally coming to me with the exact same questions and concerns and problems. So, there's no correlation there. You can never assume that someone knows how to document well or knows what you want documentation to look like. Dr. Maelisa McCaffrey: And you can, you can never assume that they're actually doing their notes either without actually checking up on that. So, I mean, that's kind of the basics, but, but then it's, it's everything around that. And. and, and then on the flip side, I often work with people who are in a group practice and are afraid to tell their supervisor or the practice owner because they know that the practice owner hasn't checked and they're terrified of when it's going to happen. Dr. Maelisa McCaffrey: And so in both scenarios, I do the same thing, which is looking at the deeper issue. So what is the reason this [00:23:00] person is behind in notes, which often, like I said, is, is not just time management. It's, it's typically time management plus. All the other things. So is it taking you half an hour to write a note and that's why you can't get them all in because that's not sustainable with 25 clients a week. Dr. Maelisa McCaffrey: Is it, is it that you literally have no time in your schedule to do notes? is it that you feel so paralyzed by the fear of not knowing what goes in a note that you avoid the task? Is it that you have a chronic health condition or, or an acute condition, like you're recovering from surgery and so, or, you have an aging parent that now you're taking care of. Dr. Maelisa McCaffrey: So those are people where maybe this has not traditionally been a struggle for them. And then something happens in their life that requires a lot of time and effort and, and maybe just logistically makes it difficult to sit in front of a computer for hours. So, you know, which can be an issue for people who have things [00:24:00] like, like chronic migraines and that's not as big a deal if you're seeing a lot of people in person, but maybe that is more likely to be a problem if you're on a screen all day long. Dr. Maelisa McCaffrey: And so then that impacts, I can be on a screen for hours talking with people, but I cannot add another admin work done. So it's, it's really identifying, like, those are all very different. problems to solve, right? So identifying the reason behind it and then addressing that issue. Meghan Beier, PhD: Yes, absolutely. And I, you know, as somebody who had a baby about a year ago, having to kind of re finagle my schedule, you know, with an infant, I really had to figure all of that out to, you know, figure out when I was going to do my notes again. Meghan Beier, PhD: Cause previously I didn't have to worry about that. I could just do it at night and couldn't do that anymore. So sometimes like big life changes can have an impact as well. [00:25:00] So I know we're talking about all the challenges. And so I think that's where AI is stepping in, right? They're saying we see people have problems with getting their notes in. Meghan Beier, PhD: We know that people are wondering, what insurance companies want. And so we can play a role here. And that sounds great. But before we dive into some of the pros and cons and ethics, from, from your perspective as a mental health professional, how, how are you defining AI? Dr. Maelisa McCaffrey: Yeah, and obviously, you know, you can look up a definition of, of AI, but in a nutshell, it is a technology that simulates human learning and synthesizes information. So it is not creating, information as far as like, it's, it is, what it is creating is from pieces of information online, literally pieces of information that it has found on the internet. And then it is kind of mixing and [00:26:00] matching things and synthesizing that information together. Dr. Maelisa McCaffrey: so it's. It's not, and it will learn and do better based on feedback you provide it, but it is also limited in that it needs to have that information and it needs to, to have a base level of information as well. Meghan Beier, PhD: Okay. and I know, from kind of watching some of your videos that you've had experience and exposure to a couple of the different AI platforms that are out there. Meghan Beier, PhD: So, you know, from your experience, both with that and talking with providers, what do you see as some of the benefits of using artificial intelligence with therapy and note writing? Why are therapists interested in it? You know, what are the advantages for them and, and even maybe for clients? Dr. Maelisa McCaffrey: Yeah, absolutely. Dr. Maelisa McCaffrey: So I think the, the biggest one that comes up, well, there are two big ones and that's time and confidence. So [00:27:00] those are the two big ones that I hear about from most people. And it's that I don't actually have to take the time to write the note. I will say in reality, once you start playing around with these systems and for people who are using them regularly, it's probably about the same amount of time writing a note using AI as it is using maybe a really good template, because you do have to, in most of these systems, they're not embedded in the EHR, so you're, you're logging into one platform and, you know, doing the note there like, or the AI is kind of, is completing the note for you and then you're having to copy and paste it into your EHR. Dr. Maelisa McCaffrey: and so there's some logistics there that, that take a little bit of time and you need to edit the note always. So, Mm-Hmm, . so that takes a little bit of time too because you're reading through it and potentially changing information. but even if the time is equal, the effort feels [00:28:00] far less and related to that is this confidence issue. So people feel like it is easier and takes less time because I don't have to do the work of putting in the clinical data. And it just comes out at me in these beautiful terms. Sometimes the clinician themselves doesn't quite understand. So we'll, we'll get to that when we talk about the negatives, but, you know, it sounds so professional. Dr. Maelisa McCaffrey: And it uses all of the right terms and it sounds so clinical and one of the most common things I hear from people is I could never write a note like that. I could never write a note that good. So it feels like it is writing this note that they could never produce on their own. Meghan Beier, PhD: And Dr. Maelisa McCaffrey: it's doing it. With with almost no effort on their part, depending on they all work a little bit differently. Dr. Maelisa McCaffrey: All of these AI platforms and for a lot of them. There are multiple ways you can give [00:29:00] it the session data. But the easiest and most common is that the AI is listening in on the session. You log into their platform and the most common would be conduct a telehealth session and the AI can then create a transcript from that and then extrapolate from the transcript, information to create a progress note. So, so that's the most common. You can also type in a summary, you can dictate a summary, you can upload a transcript, you can upload a recording. So there are other options there with, you know, depending on the platform you're using, but that's, the most integrated and simple to use because you're not really doing anything different. Dr. Maelisa McCaffrey: You're just logging into your, do your session as you normally would. And then afterwards a note is provided for you and you're just editing the note rather than producing it. Meghan Beier, PhD: Okay. And, so our practice owner has. worked with [00:30:00] AI and has quite a few patients that she's, exposed, you know, exposed and explained it to. And she said that, you know, she has quite a few who are very excited about it or think it's very cool. clearly there's also clients who maybe are concerned about it. have you heard anything about kind of client reaction to these kinds of tools? Dr. Maelisa McCaffrey: Yeah, from clinicians that I've worked with. So, you know, I don't have a practice, so I'm not using it with clients necessarily. Dr. Maelisa McCaffrey: Okay. All right. But when I've talked with clinicians, almost, I can't think of a clinician who has told me that they had clients have an issue with it. so, you know, I don't want to say a hundred percent, but I, I honestly, I can't think of someone who has said that it was a problem. clients seem to be overwhelmingly receptive to it and it's really not that big a deal for them. Dr. Maelisa McCaffrey: Interesting. Meghan Beier, PhD: Okay. You know, cause I hear a lot of concerns from therapists who are worried about, you know, how their clients might [00:31:00] react if, if they tell them that they want to start using this tool. So that's really, that's really interesting. I wanted to switch gears and talk about some of the concerns. Meghan Beier, PhD: And so I, I pulled a few questions just from some of the videos that you've posted on YouTube. so I'll start with this one in, in some of the, that work you've highlighted some kind of major concerns. one is what happens to the data and is this part of the mental health record? so can you elaborate on that one? Dr. Maelisa McCaffrey: Yeah. And I think that's, that's the biggest one that I personally have a question mark around because I think this is very much to be determined and I, I don't want any of us to be the person who potentially is in the scenario where we are the example case. because it's, you know, nobody wants to be in that position and someone will be at some point. Dr. Maelisa McCaffrey: [00:32:00] So you typically log into a separate system. You're not in your electronic health record. So regardless of how you give the, that system or that platform, the information about the session, that is very clearly PHI and that is not up for discussion. it is absolutely PHI. If you are giving a system information about session data or session information. so that's fine. If you are uploading that into a system that is HIPAA secure and you have a business associate agreement with them and, and most of them at this point are very good about that and about reminding you about those concerns and being secure. Right? So just don't use chat GPT. Dr. Maelisa McCaffrey: I have to put that out there. but, assuming you're logging into this platform for this purpose, and so everything is secure, and then you copy and you paste that into your EHR. You think, okay, well, that's [00:33:00] all good. Well, what happens with that data in this other platform? And is that considered a medical record. And if it's not, then that means you have information about that is, it is literally a transcript or a very detailed description of the client session. So then why would you have that information in any system that's not a client record? Right. So there's, it becomes this scenario where having that information housed in this AI platform has no benefit, regardless any way we look at it or tease this out, there's no reason for it to be somewhere else. Dr. Maelisa McCaffrey: And it probably shouldn't be. So what does that mean? That means then needing to delete that data in the AI platform, which to me is the easiest solution to this problem of potentially having information about a client session in two locations. [00:34:00] And I can talk a little bit more about why that potentially is a problem, because you might think, well, maybe that's not that big a deal. Dr. Maelisa McCaffrey: They're both HIPAA compliant. They're both secure. Like, what's the problem? but you have it in this other system, and some of them will delete things for you automatically on maybe a timeline and so that's ideal, but most of them don't. And, and there's all different levels of what is being deleted. Dr. Maelisa McCaffrey: So maybe it deletes the audio transcript that it recorded, but it doesn't delete your session note that you produced and saved. In order to then copy and paste it into your EHR, maybe it deletes the record, the other, the video recording of the session, or maybe it deletes the transcript, but again, saves the session note, maybe it doesn't save the session note, and then if you forget to save it, or something happens with your computer while you're in the middle of the note, now you don't have it, and so, so there is some [00:35:00] balance there of what is user friendly, what is useful, and managing having client data in two systems. Dr. Maelisa McCaffrey: So, and I will tell you, I, I have ADHD. I take medication for it. I'm officially just brought, you know, diagnosed and I am open to AI. And I can imagine I would have been one of these clinicians who's like, yeah, let me use one of these AI platforms. And I know I would struggle to remember to delete all this information. Dr. Maelisa McCaffrey: So I might think, well, oh, okay, well, I'll use it, but I'll just delete it. Like every Friday, I'll go in and delete it. And I know that would be a struggle for me. It doesn't mean it'll be a struggle for everyone. So for a lot of people, that might be a very realistic solution, but be honest with yourself about, is that a realistic solution for you? Dr. Maelisa McCaffrey: and, and then I, I also think it needs to be part of the client consent process. Meghan Beier, PhD: So Dr. Maelisa McCaffrey: when you get, informed consent from [00:36:00] people and tell them I want to use AI, which it's at this point, we have very few standards around AI as a profession, but, but that would be one of the standards is that you should get client's permission before using it, their consent. Dr. Maelisa McCaffrey: And so I do think that part of that consent process needs to be like, I will have this information housed in this other location, or I will have this information housed in this other location for a short period of time, you know, however often it's deleted, and hopefully the, the platforms themselves will address this issue by doing an automatic deletion process, and, and making that more commonplace, but it's, it's not, and it's definitely not standardized at this point. Dr. Maelisa McCaffrey: So why, why is this an issue? It's an issue because let's say, you are using a platform that does create a transcript and I will even tell you as someone who was going in and is a documentation expert and was [00:37:00] testing these and creating fake notes out of these, right? From these, these sample sessions I'm uploading. Dr. Maelisa McCaffrey: I did not think of this immediately. So, you know, I could see how you could be using this for many months and not think of this issue. I'm creating this note. Okay. And then I even think, okay, yeah, maybe I'll go in and delete that note. Well, I'm not thinking about the fact that this platform is also saving a transcript. Meghan Beier, PhD: So not Dr. Maelisa McCaffrey: just a progress note, a full transcript of the entire session. Now, let's say I haven't thought about that fact. And so now I realize, Oh my gosh, this other platform, I have all my nice, neat, wonderful progress notes in my EHR. And then in this AI platform, I have full transcripts of full therapy sessions. Dr. Maelisa McCaffrey: And I have not done anything with those. They are still in that system. And then my client requests their records or someone else requests it, or there's a subpoena and it says any and all records. I don't, I don't even know if that counts. Yeah, I [00:38:00] think that it would be very easy for someone to argue that it does. Dr. Maelisa McCaffrey: And do you want people having access to full transcripts of sessions? So I am not one to, to, encourage people, encourage clinicians to write very little and avoid details out of fear of liability, because I lean a little bit more on the other, I lean more in the middle, but maybe a little bit more on the other end of things that actually you are more likely to have, usually more information is more helpful in the long run. Dr. Maelisa McCaffrey: But I do not think it is beneficial to have transcripts for many clients of their full session. And so, you know, that's something that could potentially be very easily misinterpreted and harmful. And this therapist has never thought of that happening, right? Right. So that's why that could be a really big issue. Dr. Maelisa McCaffrey: And I, I just feel like it's bound to be an issue because as a [00:39:00] client, I might say, well, what about those transcripts? And then you as my therapist might say, well, you know, what do you, what do you mean those are in that? Well, aren't you using that AI platform, doesn't it have transcripts of our sessions or doesn't it have recording? Dr. Maelisa McCaffrey: And now we're in a position where me as the client, I am requesting information that you don't want to give me, and it is very, the way things are at this point, it would, it would be very unlikely that client does not have a right to that information. And again, that may not always be in someone's best interest. Meghan Beier, PhD: Right. Yeah. I, the subpoena thing was the thing that came to my mind at first. You know, I, I've had several situations where, for a variety of reasons, people were going through legal cases and we received, you know, requests for records and I would not want to hand over transcripts of every session, for a variety of [00:40:00] reasons just because, you know, in some, some cases it may not be helpful to the, the clients, in some cases, just too much information. Meghan Beier, PhD: so, yeah, that, that definitely is one that, I think is a major concern. Meghan Beier, PhD: you also mentioned that there were three or more kind of consent issues with some of the AI tools, in some of the videos I, I watched, so for example, with like telehealth, AI, and the recording tools, and so I'm wondering if you can say more about, what you mean by consent issues. Dr. Maelisa McCaffrey: Yeah, so number one, you know, obviously there's informed consent just with using AI, and, and, kind of ethical principle behind that is that it's new, and we're all testing it out, and all of these platforms are very new, and they are changing rapidly, and so it is very much a new frontier technology, that most [00:41:00] of us do not have a standard for. AMHCA has released a specific ethical standard about using artificial intelligence. None of the other ethics codes have, and so we literally have no guideline here other than what we would use for using any new technology or new practice. So, so there's that, just letting people know, I don't know what the ramifications might be. I don't know what this might do or not do, like, there's a little bit of unknown there. Dr. Maelisa McCaffrey: The other thing is around recording. Like you mentioned, so almost all of these platforms are doing some form of recording and in some states you might need to literally do a separate informed consent that is related to recording. Dr. Maelisa McCaffrey: So for some of us, like that is part of the law. If you're going to record a therapy session, it just has to be a separate consent form. and so know your state laws around that, as well as your own ethical guidelines. and I would say regardless of those two things. It is a standard in all of the ethics codes to at [00:42:00] least inform people, that, you know, this would include a recording. Dr. Maelisa McCaffrey: And then, and then potentially telehealth. So if you're doing a teletherapy session, so you can also do this by hitting record on on a device. in person, and it records the two of you, but, you know, the more common issue is, or, scenario is logging into a telehealth session. So then you might have just in general informed consent around using telehealth, which different states require as well. Meghan Beier, PhD: Right. Okay. and there's some additional concerns that you've brought up and these were super informative for me as I was starting to explore this, but, you had mentioned that computer programmers may not. Prioritize confidentiality, like therapists. Can you say more about that? Dr. Maelisa McCaffrey: Yeah, so I have talked with, with many of these companies. Dr. Maelisa McCaffrey: Most of the companies that I have reviewed on YouTube, I have actually talked with the founders [00:43:00] of, of those companies. And so I will say that most of them seem very well meaning and, and truly have like a very personal story of starting this or started in the medical field in some way and found out through their research, how big a deal mental health issues were and wanted to help therapists. Meghan Beier, PhD: So, Dr. Maelisa McCaffrey: you know, I do not see a lot of these companies as like big bad tech companies and they are not clinicians. The vast majority of people starting these companies are not clinicians. They are people who have an expertise in this technology, and see an opportunity that can potentially be both profitable and helpful. Dr. Maelisa McCaffrey: and so, so they have really great intentions, but because of that, they're not thinking through, for example, the recording issue. They're thinking, this would be awesome, this would be so helpful. And they're not thinking through it may not be helpful to actually have full transcripts housed in a platform [00:44:00] full transcripts of therapy sessions, right? Dr. Maelisa McCaffrey: And so for many of them, even when I have talked with them just to get information about their platform and learn about it, you know, I might mention something and they're like, oh, and they're taking notes and, and starting to ask these different questions. Another thing that they often don't think about is insurance. Dr. Maelisa McCaffrey: And so I would say that's, that's becoming more common now, but especially like even a year ago when I was doing a lot of this, most of them were not thinking through what does insurance want to see in a progress note and making sure that that's included. So they just have very different, you know, they're coming at it from a problem solving perspective. Dr. Maelisa McCaffrey: And a technology can be very helpful perspective, which is often very different from how therapists approach things like most therapists tend to be very wary of new technologies. Dr. Maelisa McCaffrey: so there's a big clash there a lot of the [00:45:00] time. And what it leads to then is therapists being a little overly cautious and communicating with these tech companies who are on the other end of that spectrum. Dr. Maelisa McCaffrey: And so then they kind of don't really know what they should actually be cautious about. And then they need to seek their own consultation. And so, you know, it's, it's just such a different approach. And because the communication can be somewhat poor, as far as, You know, what is, what is an ethical standard and because a lot of this stuff is not very clear cut either because it is so new, then they, they have very little guidance other than HIPAA. Dr. Maelisa McCaffrey: So they, they know they need a business associate agreement because they're going to be a technology platform working in a mental health field, managing PHI. So that's good. That's a good baseline. but even with that, you know, there have been issues with EHRs recently, [00:46:00] may know that. On a baseline level, but don't understand the deeper concepts of confidentiality and and the ethics of it and what that can look like and the reasons why we have such strict guidelines. Dr. Maelisa McCaffrey: So, so there's, there's a lot of education that needs to take place. And some of these tech companies are better about seeking that out than others. Meghan Beier, PhD: Yeah, makes sense. you had mentioned earlier that people should stay away from ChatGPT, and we mentioned HIPAA compliance. Can you say more about that? Dr. Maelisa McCaffrey: Absolutely. So, ChatGPT is not HIPAA compliant. even if you pay for it, you just get a better version of ChatGPT if you pay for it. So, ChatGPT, Gemini, I also like Gemini a lot. Like any, anything that does not give you a business associate agreement is not, means that you are not being HIPAA compliant. Dr. Maelisa McCaffrey: So, even if a [00:47:00] technology tells you it is HIPAA compliant and has a nice little like HIPAA symbol on their website, if they don't give you a BAA, you are not being HIPAA compliant. So that's really key to know you need to have that. and so ask them before you enter any client data into any system. Dr. Maelisa McCaffrey: So then the thing that comes up with chat GPT and Gemini, because it's free and these other platforms that are HIPAA compliant, typically you might get some kind of free trial, but it's not free. You're paying for this extra level of protection, and that's what I would encourage, how I would encourage people to view it. Dr. Maelisa McCaffrey: so the free platforms are, are better in some ways because they're way, they're all way faster, like dramatically faster, than the paid platforms at this point in time. so it can feel like it's almost better. But they are not. And they will tell you if you try to enter a fake progress note in the chat GPT, it will give you a notice and tell you we this is not a place to enter [00:48:00] PHI protected health information because they are not securing it. Dr. Maelisa McCaffrey: It is open source and Gemini and ChatGPT and a lot of other AI platforms are using information you put in to continue to teach its system and its platform. And they are using that information for a whole host of things, so it is absolutely not secure. And if you think that you are entering information into ChatGPT and hiding PHI. Dr. Maelisa McCaffrey: So a lot of people will tell me, well, I'm going to write a progress note using chat GPT, but I don't put enough information in there for it to be identifying. And the consistent feedback I've gotten from other people who are in our field, who are both in the legal aspect and in the ethical aspect, is that, and I would agree, it is impossible to create an individual progress note for an individual client without entering [00:49:00] PHI. Dr. Maelisa McCaffrey: So it is either a generic note that I could write for anybody and that I could just make up that has nothing to do with this individual session or it is a note that is specific to this individual client and therefore I had to enter information specific to that client that could be considered PHI because PHI is not just name, date of birth, insurance number. Dr. Maelisa McCaffrey: It is anything that could be identifying of the client. And the combination of factors, it doesn't have to be 1 thing specifically. It can be a combination of information that you include. So, and the fact that based on our ethical guidelines. There are lots, I mean, I can't believe how many options there are. Dr. Maelisa McCaffrey: There are lots and lots of AI options that are specifically for mental health therapists to write progress notes and treatment plans. So you cannot argue that there are no better options out there. There are lots of paid options that agree to protect your client's [00:50:00] data. So there is absolutely no reason to use ChatGPT or Gemini. Dr. Maelisa McCaffrey: For individual client information. Meghan Beier, PhD: If somebody has been doing that and they're starting to panic listening to this, what would you say to them? Dr. Maelisa McCaffrey: I would say just stop as of today. Yeah. There's nothing you can do about, you know, formally using it. I wouldn't recommend it. I would, I would literally stop today and check out some of the paid platforms. Dr. Maelisa McCaffrey: Because if you've liked using it. The benefit of the fact that you've tested it so far is that you kind of know you like using AI for progress notes. And so you'll probably have a much better idea going in to check out different platforms of what you're looking for and how you want to be able to use it. Meghan Beier, PhD: Okay. Sounds good. I, yeah, I wanted to also, you know, I haven't seen this in any of your videos, but maybe you have talked about it, but do you, do you have any information about [00:51:00] how AI might be biased against marginalized groups? Dr. Maelisa McCaffrey: Yeah, so what I have learned is that it's basically that most programmers are not indicative of the general population. Dr. Maelisa McCaffrey: So people who are programmers are not necessarily like they tend to be overwhelmingly more male, for example, or overwhelmingly from one or just a couple of racial groups. they are not representing the population and the same way that the general population exists. So in that way, there are just, as we all know from bias training, We can't even know what all of our own biases are, and so if you are literally teaching something to learn, then you're not even aware of the bias you are inputting. Dr. Maelisa McCaffrey: So, a common example that people may have heard of is facial recognition. So facial recognition works a lot better on [00:52:00] people who look Caucasian than it does on people who don't. and, and there's a reason for that. It's because the information, it doesn't mean that like that type of technology can only work on people who have whiter skin. Dr. Maelisa McCaffrey: It just means that it has only been taught to work better on people who have whiter skin. So that is one way where, you know, it does highlight that it's not this. I think we, we overemphasize the intelligence part of artificial intelligence because you would think it could extrapolate. And that the color of someone's facial features shouldn't matter. Dr. Maelisa McCaffrey: but it does. And that's because it really is relying on information it has from the internet or from the, the, you know, companies and people who are training it. So, it's, it's limited. It's very, very limited. And, and that will be the case for quite some time. It is something a lot of companies are actively working on and [00:53:00] want to improve, but that is the reality still today. Meghan Beier, PhD: Right. Right. And as, as somebody who works in the health psychology field or people with disabilities, you know, there's probably a lot of ableist language that's popping up in some of these things too. at least some of the, the, the one that we've kind of, tested out in our practice. I was noticing some of that too. Meghan Beier, PhD: Yeah, absolutely. Yeah. Meghan Beier, PhD: so I know we've talked a lot about some of the concerns, but as we're sort of nearing the end of our discussion today, I want to switch to, you know, there, there are some pros. You know, you mentioned that you have ADHD, similarly, I, I also experience executive challenges and, you know, some, some, language based writing challenges. Meghan Beier, PhD: And so AI has been of interest to me, for those reasons. And so, you know, I guess I'm wondering, for those who do want to do this and want to do it ethically, you've kind [00:54:00] of looked at some of the tools on the market and I, you know, I'm not asking you to sort of say like advocate for one tool over the other, but I would say, you know, kind of what are the pros and cons? Meghan Beier, PhD: What, what have you found, seems like something that could be something to look for? And, you know, any, any ethical obligations that people should be keeping in mind as they're thinking about which tool to choose. You've mentioned quite a few already, but, anything to add? Yeah. Dr. Maelisa McCaffrey: Yeah, I think for me, the, the biggest benefit that I've heard from people, it goes back to what I was talking about earlier with mental effort. Dr. Maelisa McCaffrey: So even if you take the same amount of time to complete your notes total, if the mental effort is so much easier, because for a lot of people, I can absolutely understand how it's easier to go in and edit a note than it is to produce a note. Even though, like, I try to give people all these wonderful templates that they can use so they're not having to think [00:55:00] about what to write. Dr. Maelisa McCaffrey: And yet, I still tell people, I can only do so much. You know, you, you will have to adapt this template to your needs. and so, AI cuts out a little bit of that because it's, it's only recording on that session. and so, it's really helpful. So I've heard people Say I tested AI platforms and they don't work for me and it seems to be a hassle and I've heard people, I've heard multiple clinicians use the phrase life changing using AI has been life changing for me because it has saved me so much time and effort. Dr. Maelisa McCaffrey: So, so I think that's the, the, the key there. If you are someone who continues to struggle despite attending lots of trainings and, and training. And working on your template and creating all of these personalized resources, and you're still not confident in your notes. If AI can give you that confidence, I, you know, I say go for it. Dr. Maelisa McCaffrey: I am all for whatever gets your [00:56:00] progress notes done and gets them done more easily. I think with that, the key is to know what to look for when you're looking at these progress notes and do know that for one, you know, they all take longer to write than you would ever imagine. So it is literally a matter of like, go do a second session while the progress note is being written for you in the background and then come back to it. Dr. Maelisa McCaffrey: So you can't, you have to do a little arranging of your schedule in that way, because you will have to come back to the note, you know, to review it. And it's not going to be there at the end of your session, you're going to have to hit some button or wait 20 minutes to see it. So if that's what you're expecting, it's not going to be the immediate gratification, but it will initiate the task. Dr. Maelisa McCaffrey: And for a lot of people that, that initiation of the progress note is really the biggest barrier. And so if that's you, then totally try AI because it takes that step out. And so you might be more likely to log in and say, Oh [00:57:00] yeah, what did it write? And then I can just quickly edit it and the big thing there is. Dr. Maelisa McCaffrey: If you are working with insurance, make sure that information makes sense with the treatment goals, make sure that information is highlighting the client's progress and their ongoing need. I am, in two sentences, distilling what could be an all day training. So, you know, there's that, but, but know that it, you know, do, you do need to check that it's including the information that insurance would want to see. Dr. Maelisa McCaffrey: And, and know that every platform I've tested has made things up including client quotes. So you do have to read the note and edit it potentially. so don't skip that step because you would absolutely be mortified if you misquoted somebody. I mean, these are really obvious things, you know, so most of us would be mortified if that happened. Dr. Maelisa McCaffrey: So do read the note and edit the note. but if, if the, the constant doubt is a struggle for you, and if the initiation of getting notes done is a [00:58:00] struggle, AI will probably be helpful. And just like with finding an EHR, you probably need to test out a couple, and I would say test them out based on how you think you will give it information. Dr. Maelisa McCaffrey: So if you do a lot of telehealth and you think you're going to use the recording feature where it's just listening in on the background and then producing the note afterwards, which is kind of the easiest to do, you know, then test platforms that use that because they don't all have that feature, versus if you think you're going to be using dictation and kind of dictating a summary afterwards, make sure you look for platforms where that is a really easy, and, and, and prominent feature of the platform. Meghan Beier, PhD: Excellent. And last question. if our listeners want to learn more about the trainings you do, or want to learn more about you and, and your educational resources, can you share a little bit about where to find [00:59:00] that information? Dr. Maelisa McCaffrey: Yeah, absolutely. So you can go to QAprep.com. com and that's my business QA prep. Dr. Maelisa McCaffrey: everything is there as far as all the, I have online courses. I have a book coming out in October, which is really exciting. And So, you know, I have templates, all the things related to documentation. And if you are interested in, AI platforms, I have not reviewed all of them. Unfortunately, I've had to focus on my book a little bit more recently, and there are lots of reviews in the queue. Dr. Maelisa McCaffrey: So depending on when you listen to this, I may or may not have more platform reviews. but that walks you through, what each of these platforms offers, as well as how well they all wrote a progress note for the exact same session that I give all of them. So it helps you, compare and contrast a little bit. Dr. Maelisa McCaffrey: And, you know, that might be a good place to start. So, my YouTube channel is Maelisa McCaffrey, and my first name is spelled a little oddly, but if you put in Maelisa McCaffrey, you will, [01:00:00] or if you put in QA prep AI, anything like that into YouTube, you will come across those, those videos. And there's a whole playlist there that you can watch. Meghan Beier, PhD: Excellent. And I'll put all that in the show notes so people can find you. Thank you so much for this discussion. It was really helpful. Lots of really great ethical nuggets to learn from. I really appreciate your time. Dr. Maelisa McCaffrey: Yeah. Thank you so much for having me. Conclusion --- Meghan Beier, PhD (2): If you like our show, please like subscribe or leave a review. This helps others find out about us. Find Empathy provides continuing education. If you would like to learn more about our courses, please click the link in the show notes or visit learn.findempathy.com. While our education is free, for example, like listening to this podcast, the proceeds we receive from continuing education goes to support the maintenance of our website and the development of new content. Meghan Beier, PhD (2): Finally, please know that the opinions expressed by the experts today are their own. We are not financially supported by any of the businesses or resources [01:01:00] described in today's podcast. Additionally, any advertising that is published on the episode is chosen by our podcast hosting service and not by Find Empathy. The content provided today is for educational purposes only. Meghan Beier, PhD (2): Thank you so much for listening and we look forward to you joining us in the next episode.