Unknown (4:41)

Can you give me a hypothetical of.

Dena Temple Rouston (4:43)

How it would work?

Dr. Stephen Xenakis (4:44)

Well, the hypothetical would be, and we have tested this out, is that a patient comes in, let's say a veteran. And we know that these veterans have multiple problems. They've got ptsd, many have been exposed to trauma, and they have traumatic brain injury. They have had other injuries, so they've got chronic pain. They may have sleep problems. They may have been exposed to medications that have also made them feel more anxious or caused thinking problems. And they come in for an interview and they say, look, I need some help here.

Dena Temple Rouston (5:21)

So how do you figure out what this person needs? The traditional response was use a questionnaire, fill in some bubbles, rate your mood on a scale of 1 to 10, things like that. But Dr. Zenakis says, no, that's just not enough.

Dr. Stephen Xenakis (5:37)

We're getting away from questionnaires. That's what I'm trying not to do. I think questionnaires, excuse me, I think are too crude. That's standard in the field. But I don't think they get the kind of nuanced subtleties that we get just from an interview.

Dena Temple Rouston (5:55)

What Dr. Xenakis would do instead of multiple choice questions, is a more open ended kind of interview with the patient.

Dr. Stephen Xenakis (6:03)

It would just be a raw interview that could be uploaded and compared to what we think is our reference interview or our sentinel. And how close is the correspondence? And because of the way these large language models work, they're able to find key words and nests of words that show what the correlation is. So this goes way beyond, it's quantum leaps beyond a questionnaire.

Dena Temple Rouston (6:31)

So this is the idea, not just forms, not just checklists, but something closer to the way real conversations work. And then AI would help doctors scan those conversations and pick up on things that might otherwise get missed. Maybe just some nuance that got lost in the shuffle of a long conversation, or it could be something that a psychiatrist just doesn't have expertise in, but the AI has been programmed to understand. Dr. Zanakis says his interviews with patients would be structured in a way to actually help the AI.

Unknown (7:06)

So would there be set questions so that you could compare apples and apples?

Dr. Stephen Xenakis (7:11)

Just a good clinical interview? I mean, I think good clinical interviews have set questions. I mean, what is the problem? How serious is it? How long have you had it? You know, what other problems do you have like this? I mean, what's the history have you had? I mean, it's just a good clinical interview, right?

Unknown (7:28)

Are you sleeping? Which is usually a pretty good one.

Dr. Stephen Xenakis (7:31)

Yeah, exactly.

Unknown (7:32)

To tell if something's off. I see.

Dena Temple Rouston (7:40)

Pairing these kinds of structured interviews with AI, he says, could take the doctor patient relationship out of multiple choice and into something he thinks would be, well, deeper. Which raises the question, could technology, something that feels cold and impersonal, actually help make mental health care more human? And do we even want it to stay with us?

Matt Galloway (8:13)

When was the last time you said, hmm, I never thought about it that way? The Current aims to give you that moment every single day. Hello, I'm Matt Galloway, and our award winning team brings you stories and convers to expand your worldview. Sometimes they connect to the news of the day, sometimes to the issues of our time, and you'll hear all kinds of people on the Current, from bestselling authors to maybe your neighbor. Find us wherever you get your podcasts now, including YouTube. I'll talk to you soon.

Dena Temple Rouston (8:52)

So imagine you're a psychiatrist in a small town, maybe in Wyoming, maybe Alaska. You've got veterans coming into your office, and some are struggling with ptsd, others with pain that just won't go away. And they need your help. But you don't have a huge team. Instead, you have a backlog of patients and not a lot of resources to go around. What Dr. Xenakis hopes AI can do is give doctors access to something that could both extend their expertise and make the treatment process more efficient by connecting doctors to a system that could take every interview, every conversation they've had, and compare it to thousands, maybe hundreds of thousands of other conversations.

Unknown (9:41)

Are there particular places where you think this would be particularly helpful?

Dr. Stephen Xenakis (9:45)

I mean, look, I think it'd be particularly helpful outside of our urban areas, places where we don't have a good number of mental health people, and yet we have a number of patients that are really seeking help in the military.

Dena Temple Rouston (9:58)

There's a term for this kind of thing. It's called a force multiplier. A Force multiplier is anything, a weapon or a tool that lets soldiers do more than they could do on their own. It essentially amplifies their abilities. And Dr. Xenakis, he says AI could be a force multiplier for doctors.

Dr. Stephen Xenakis (10:18)

I know that there are many therapists out there that are in communities and may not have had a lot of experience.

Dena Temple Rouston (10:25)

And the numbers are kind of shocking. In states like Alaska and Wyoming, veterans make up at least 10% of the population. But the number of psychiatrists in these states, well, for the roughly 1.3 million people, there are less than 200 registered psychiatrists in both Alaska and Wyoming combined. Psychiatrists in rural areas are strapped not just for resources, but for patient data to refer to. But a large language model could essentially expand the amount of patient data they can draw from to help guide their patient care.

Unknown (11:10)

And then what they would be able to do is, with a large language model, it would be compared with thousands, maybe hundreds of thousands of other patients. So that you'd be using all that knowledge to come up with a solution.

Dr. Stephen Xenakis (11:22)

Yeah, exactly. And then it would learn over time, so we would be able to feed that back. The more we do that, the more we learn. And we say that, for example, a patient's prescribed a particular medication that we think would help with sleep. Well, as we look at that and look at those patients who were prescribed and look at their interviews, we can get a better idea of who would be a responder. And that's really important.

Unknown (11:51)

Right, so that they don't take a.

Dena Temple Rouston (11:52)

Pill that doesn't end up helping their problem.

Dr. Stephen Xenakis (11:54)

Yes, right.

Dena Temple Rouston (11:55)

Because even with a great interview, even with the best doctors, no one gets everything right. But maybe with enough data, they could get closer. So let's be honest. Anyone who's ever gone to a doctor has probably fudged a thing or two. You say you exercise a few times a week, but really, was it twice? Once, not at all. You tell your doctor you sleep okay, but if they asked you how many hours, how much was deep sleep versus light sleep, you wouldn't really know. Xanakis knows this. He knows that some of the information he gets from his patients isn't exactly, well, precise.

Dr. Stephen Xenakis (12:47)

It's really hard to get good information from your patients about how well they slept or really did they exercise? Not that any of this intentionally being misrepresented or, you know, that they're not intentionally not disclosing to us. I mean, how well did you sleep? I don't know.

Dena Temple Rouston (13:09)

But now we don't have to guess about some of those things because we have wearables. So think about something like an Apple Watch or a Fitbit. It tracks your heart rate. It knows how many steps you take. It even knows down to the minute how much deep sleep you got last night. And when doctors have access to that kind of data, it changes the game.

Dr. Stephen Xenakis (13:30)

So let's look at that and then let's look at what you did, or let's look at your diet. And the more we have that and we can look at it together, the better decisions we can make. And then we can decide what treatments might work best with you as we think about either medicines or activities or cognitive behavior therapy.

Dena Temple Rouston (13:51)

Because mental health isn't one size fits all. What works for one person might not work for another. But AI, these wearable devices and the technology being developed now, Dr. Sinakis says this is a real turning point, a way to customize treatment better and get past the guesswork.

Unknown (14:10)

So I just got a new Apple Watch. And so I have it tracking my sleep because I was interested. I'm not really interested in all the exercise stuff and heart rate and all that other stuff, but I'm finding myself being quite obsessed with how much deep sleep versus core sleep versus REM sleep, how much I'm awake. And it is, as you say, something.

Dena Temple Rouston (14:34)

I would never be able to tell a doctor.

Unknown (14:36)

I could just say, okay, I feel pretty rested, I guess, but this is showing me just how much it is.

Dr. Stephen Xenakis (14:41)

Yeah. And I think it's very. I mean, I know that people get obsessed with it, but I think it's really helpful. I use it a lot with my.

Dena Temple Rouston (14:51)

Patients using technology in psychiatry. It sounds futuristic, maybe even revolutionary. The rub is that psychiatry isn't an equation. It isn't about solving for X. It's about people. And people can be messy. And then there's this bigger question. Is this what we really want? Sure, AI has its place. It can help with paperwork, maybe spot early warning signs. But psychiatry isn't just science. It's an art. And AI isn't an artist. At least not yet. From recorded future news, this has been Click Here's Mic Drop. It was written and produced by Erica Gajda, Megan Dietri, Sean Powers and me, Dina Temple. Rest. It was edited by Karen Duffett. We'll be back on Tuesday with an all new episode of Click Here. Have a great weekend.

Unknown (16:01)

If you're looking for a daily guide to cybersecurity news and policy, sign up for the Cyber Daily from Recorded Future News. It serves up today's most interesting and important cyber stories from our sister publication, the Record, and then aggregates all of the big cyber stories you might have missed from news outlets around the world. Just go to the Record Media and click on Cyber Daily to get all you need to know about the world of cybersecurity right in your inbox.