In this episode, basic information about AI and mental health apps. Welcome to the Anxiety Road Podcast. This is the involuntary journey, finding treatment options and resources for people that have an anxiety condition or a disorder. This podcast is treatment agnostic.
My name is Gena Haskett there are a awful lot of words on the potential of AI for mental health services, on tech and business websites, and on those websites that deal with investments or speculative investments, you'll find a great deal more. I have been looking for quite some time on grassroots mental health information or mental health education information specific about apps, and I'm not finding a lot of information that is slowly changing, but not as much as I'd like.
So I think it's time that we start talking about, talking about this particular topic. And the reason being I want to talk about it is that there aren't enough mental health providers. Many of the more experienced providers are reaching retirement age and they're leaving the profession. There is a financial access problem for many of us trying to get mental health care.
For some of us, particularly in the rural sections of the United States, there are geographic and transportation access problems to obtaining a mental health service or even finding a provider. There's some areas in the United States especially in the rural areas where a person would have to drive almost 400 miles to find the nearest accessible mental health provider. And that person is probably booked three to six months in advance.
By the way, this is the podcast sharing the medical, behavioral health, meditation, relaxation, and X equals the unknown treatment options and resources for those of us on the anxiety condition or disorder spectrum. So that's the topic of this episode. I wanna give an overview of what AI is and examples of the types of apps that are either in the marketplace or due to coming to the marketplace within the next year or so. Now, this is not going to be a rah rah or hell no type of episode.
I really just wanna provide basic level information so that people can make an informed choice. But first in the news, there's a really good article by Rose Cartwright in The Guardian, it's called, I Was the Poster Child for OCD, and then I began to question everything I've been told about mental illness. Rose talks about her experiences having OCD, but more important to me, the evolving view of what is missing as part of the treatment process.
Perhaps maybe we shouldn't have every single thing, on the medical side, meaning, yes, we do need to access doctors. Yes, we need access to medication, but for a lot of us, we need access to housing. We need environmental safety. We need to have resources in the community that we can turn to in order to maintain our mental and physical health. Part of that is socialization.
And with the advent of everybody going digital, there has been a serious reduction in the amount of actual physical contact with other people. So. I think it's a very good article to take a look at and to think about some of the topics that she's bringing up. Longtime listeners know I often use the term entrepreneurial mental health. In my opinion, the focus of these businesses is to make money by delivering cost efficient mental health services to generate profits.
These services are often marketed to insurance companies, hospital systems, and large corporations. There are some apps that are directed to consumers, however, the main focus of the business is acquiring customers and how quickly they can extract money from them. Some of the entrepreneurial mental health providers have problems with concepts like ethics, privacy, and protected health information.
I commend to your attention an article posted on the MedPage Today website entitled The Dark Side of AI in Mental Health, and the author of the article gives that a very interesting, example. Basically, a company posted on Craigslist an ad that requested recordings of therapy sessions for $50 a pop. The recordings had to include both the patient and the therapist and last 45 minutes, and yes, that is cringe-worthy.
It certainly violates a boatload of ethics, privacy, and protected health concerns. Don't waste your time looking for it, because both the ad and the website have been removed from public view, probably due to the many, many people objecting to the idea concept and practice of purchasing therapy sessions and then wanting to use those recordings to train generative AI systems. So let's start with a very excessively simplified definition of what AI is.
For those of you who are on the tech bro or the tech gal persuasion, you may not find my definition according to Hoyle. That's okay. I am trying to give a conceptual understanding of what's going on as opposed to a technical, precise one. So. I want folks to think about a dictionary book, a physical dictionary. Now that book contains words with definitions, sometimes images, sometimes a pronunciation key or additional information.
It is a container of information that can only be accessed by people who can read and read the language that the dictionary is written in. Dictionary books are static fixed containers. They require direct human access. In 2024 we have software programs and apps that collect information and stores that information in a language data container. We are talking trillions of words and other content. typically this is called a large language model.
Now it holds onto that information until a human being ask a question or request information. So let's start with a really basic question. What color is the sky? The programming would let it know that the word what starts in inquiry sentence. The other words in the sentence is the are not important at this time. Color would be a thing. Sky would be listed as a thing. It then would look through the collection of data and find any information that matches those search terms.
Then it would look for any patterns that include all three of those words and any documents that would contain those words, and possibly even the question mark, there would be additional filtering processes going on to get as close as possible to the original question. And once it has the original question, it will also locate the sources and then provide the answer. The sky is blue. There was no actual thought process.
It was more like pattern matching with some advanced mathematics checked in 2024. This is what most people are talking about when they're talking about artificial information processing. So again, a very simplified definition. Now that you have that in your mind, we can move on to the types of mental health apps and what else is going on. One type of mental health app is called a scripted chat bot. A chat bot is a computer program that appears to imitate human conversations.
There is a body of knowledge or a collection of information relating to a specific topic. There can be a script or a template that takes what the user types and links it to a specific script response. So if a user types, I'm feeling blue, the chat bot does not know anything about blue the color, but in the specific large language model for a mental health app, it would look for anything that has the word blue The word blue would be mapped to a script, possibly about sadness or depression.
The AI chatbot might respond with, it seems like you're feeling sad or depressed. Again, no intelligence is being used. It's mapping to the words feeling blue with an approved scripted response. It then could offer some suggestions, again, scripted on meditation or activities to help the user feel better. Current examples of AI chatbots are Woebot and Wysa. Next up, you would have generative ai, mental health chatbots.
There still would be a script, a very detailed one with psychological and behavioral health treatment information. There's also information about how a user or a patient would probably express themselves and their feelings, and that would also be scripted. And again, as the prior example, there would be a list of types of responses that a human being would use in order to communicate about how they felt.
There would be a list of the type of treatment resources that could be combined with all of the above information. , That doesn't really exist yet. There are hybrid models where it will do a little bit more than the standard chat bot. Or maybe if it got too complicated or the chatbot didn't understand you.
It would hand you over to a human person., but a totally self-contained, no human intervention, mental health chatbot that would basically treat you like they would treat a therapist would treat you. That doesn't quite exist yet. However. There is a new app called Therabot that is in testing at Dartmouth University at the time of this recording. Maybe in about two or three years they would release it to the public.
Or an insurance company who wants to save money and doesn't wanna pay therapist anymore, and they subscribe to the Therabot service and then they would allow you access to Therabot. I have no idea how this is going to work once this gets implemented. So what are the pros of using an AI mental health app? Well, you certainly would have 24 hour access.
It could be a supplement to traditional therapy so that when you couldn't get ahold of your regular therapist, you could access the features of the chatbot. Certainly for symptom relief. It is a great opportunity for education. It is a way of distress and some of the, the chat bots that are out there have games or activities to help you de-stress any a variety of ways.
You could have progress alerts, you could have journaling, you could have some tracking functions, probably tracking your sweeping, so you could really have quite a nice comprehensive. Monitoring of you and your symptoms, and you could actually see the improvement as time goes on. What are the cons? They are significant. First up for me would be misdiagnosis. Many people will turn to an app before even thinking about going to the doctor.
There could be a physical problem, but users won't know it or, and this does happen you have an anxiety disorder and a physical health problem. Over dependence on the app. There will be people that will use the app 24 7 and ignore all other treatment options. You can't just use an app and continue to slug down coffee and three donuts a day. That's not how this treatment process works. We're talking about a holistic process where you may have to change how you eat.
You may need to include more exercise. You may need to be in a place that supports your spirit or do things that supports your spirit. You just can't rely on an app to get you through every moment of the day. Privacy issues and abuse of patient data. We have current and former court cases concerning the selling of user data to advertisers via online mental health and social media providers. For sure. You can count on a mental health app to do the same.
If you don't read the fine print of what's going on in the terms of service, many times they will tell you that they will sell your information. Other times they will probably do it with or without your permission. And dangerous advice. One specific example that did happen was there was a eating disorder chatbot, and in testing it was fine.
In initial patient release, it was fine, but another company had access to the app, made some minor changes , and the app turned around and made suggestions to people who had an eating disorder that they could provide ways of helping them lose weight or restrict their calorie intake. Something that you do not want. to suggest to people who have an eating disorder, but let me give you an even darker example. Hypothetically, what if a user was in a dark place?
It turns to the chat bot and says that they feel like dying. One of the things you do not want to happen is for the chat bot to list effective ways of off planet themselves. That would be not be a great thing. These are just the basic core fundamentals, and one of the ways to protect yourself is by information and being able to separate noise from hype, from fear, and really understanding what's going on. As always, I do have some recommendations and resources for you.
Certainly, you should check out The Guardian. That article from Rose Cartwright, the Dark Side of AI and Mental Health, and the subtitle is High Demand for AI Training Data may increase Unethical Practices in Collecting Patient Data. This is on the med page today website. The New Yorker magazine has an article called Can AI Treat Mental Illness. And if I find some more resources, I will certainly put them on the Anxiety Road podcast blog.
So that's all that I have for you today and go forth and don't fear the AI just yet. You can visit anxiety road podcast.com for links and other information sources. You can listen, download, or follow the podcast via number of iOS and Android podcasting applications. You can find it on Apple Podcast, Amazon Music, Ghana, Pandora, or on the web@listennotes.com. Links to other sites are provided for information purposes only and do not constitute endorsements.
Always seek the advice of a qualified healthcare provider with questions you may have regarding a medical or mental health disorder. This podcast is intended for information and education purposes only, and nothing in the program is intended to be a substitute for professional, psychological, psychiatric, or medical advice, diagnosis, or treatment.