a day ago
S3 Episode 5: Smartphones and the Future of Psoriasis Care
This transcript has been edited for clarity. For more episodes, download the Medscape app or subscribe to the podcast on Apple Podcasts, Spotify, or your preferred podcast provider.
Steven Feldman, MD, PhD: Hi, I am Dr Steve Feldman. Welcome to Medscape's InDiscussion series on psoriasis. Today, in episode five of season three, we'll discuss digital assessment tools for psoriasis and psoriatic arthritis with the director of both the Penn Psoriatic Arthritis and Spondyloarthritis Program and the Penn Center for Clinical Epidemiology and Biostatistics, an associate professor of medicine and epidemiology at the Perelman School of Medicine of the University of Pennsylvania, her research includes the development and validation of digital assessment tools such as smartphone applications to remotely measure cutaneous and musculoskeletal signs of psoriatic disease. Let's welcome Dr Alexis Ogdie to InDiscussion .
Alexis Ogdie, MD, MSCE: Thanks so much for having me today.
Feldman: I'm so glad you could be on. Alexis, I understand the Penn Center for Clinical Epidemiology and Biostatistics, which you direct, has about 200 faculty, including at least two of the top dermatologic epidemiologists in the world. Maybe the two top dermatologic epidemiologists. Being the director at such an important center must be a gratifying and uniformly uplifting experience. Does directing something like that require attendance at a lot of meetings?
Ogdie: It does require a lot of meetings — many more meetings than anticipated. But it is also incredibly exciting and fun, and uplifting is a good word for it because I get to hear about all kinds of things going on in all kinds of domains of research, especially clinical research and epidemiology: all the cool things that biostatisticians and informaticians are doing to make the world an easier place to study things and to kind of figure things out and watch people as they go through clinical care too.
Feldman: I'm trained as a biochemist and have no interest in that. I'm interested in health services research and the stuff Penn does in dermatology, and you know, in general in medicine. It's just amazing.
Ogdie: There's so many cool intersections between health services, clinical epidemiology, biostatistics, and bioinformatics that can interact with health services too. Adherence, which I know you study, is one of the things that sometimes takes some different algorithms to measure, too. There are lots of fun things in that space.
Feldman: One of the things I've struggled with in adherence is when I do studies that use electronic monitors to measure people's use of medicine every day. A lot of times, they're not using the medicine. It could be because they're suffering, and they don't use it. Or it could be that their skin disease went away, in which case I don't want them putting the topical steroid on it.
I need to have a measure of their disease severity all the time, because if all I have is when they come back in a month, the ones who used the medicine really well, it cleared up, and then they stopped using the medicine — it looks like poor adherence is associated with better outcomes.
Ogdie: This is one of my fun pieces of being an epidemiologist is that you get to think about how these different variables interact and the potential bias that could be causing these spurious outcomes. That's a great example of one fun area, but also talking about digital biomarkers, a great opportunity, because some of these digital biomarkers and other ways of getting data that's happening in an everyday way can help with understanding those relationships. If we only see the patient at certain intervals and measure only a couple of variables that are most important to us, then we're only capturing what we see as opposed to all the things we don't see.
Feldman: I enjoyed reading your article, 'Clinical Validation of Digitally acquired Clinical Data and Machine Learning Models for Remote Measurement of Psoriasis and Psoriatic Arthritis'. You had me at remote measurement of psoriasis. You're probably more interested in the remote measurement of the arthritis. The article begins by saying psoriatic disease remains underdiagnosed and undertreated.
Is the technology you're developing going to address unmet needs like those? Will it make the diagnosis? Is it going to recommend treatments?
Ogdie: Well, so that's a good question. So, the app that is being discussed is called the Psorcast app. So this was developed by Psoriasis and Psoriatic Arthritis Clinics Multicenter Advancement Network Consortium (PPACMAN), an organization that brings together dermatologists and rheumatologists to think about multidisciplinary care. And in that sense, one of the reasons for calling it Psorcast is, the ideal biomarker or digital biomarker would be able to say among patients with psoriasis, "Hey, you may be developing psoriatic arthritis because we notice that you're walking differently." Or we notice that your wrist movements have changed, for example, and can that help us predict or forecast who's developing psoriatic arthritis and needs to be seen by a rheumatologist?
In that way it may help us diagnose, but not the tool itself, diagnosing, but the tool saying, 'You need to get to someone who can measure that appropriately and see if we can come up with a diagnosis.'
Feldman: Tell me what does this tool measure exactly?
Ogdie: The tool measures the body surface area of psoriasis. There's a psoriasis draw tool where they can draw in where the psoriasis is. They can take pictures of their psoriasis, look at the erythema and duration, and scaling, for example. They can take pictures of their fingers, so we can see dactylitis by measuring one nail vs the other nails.
We have specific nail pictures where they can take a picture of the nails. Those are hard because obviously you can't see under the nails for pieces of it, but you can see general things there. Some of the more fun or unique measures include patient-reported outcomes, but I think the fun and unique measures that we have, there is a walking test, so we can look at the amplitude of your walk. You put the phone in your pocket, you first have to push on or go to start the test. And then it gives people instructions verbally so that they can hear it as they're doing a little walk test, kind of walking back and forth. And it's measuring the amplitude of someone's walk and we're looking for changes over time; are there changes in the walk? Additionally, it measures wrist, mobility, and it's just kind of like wrist, elbow, and shoulder. So their phone is on the table, and you're turning it, and it can measure how far you can turn it, for example. These are some different tools that we've built thinking that these may be helpful. And that was part of the study was just doing an initial validation of can we capture the data? There's a larger validation study that's ongoing to see can we measure all pieces of this? And does that work for psoriasis and psoriatic arthritis patients, too?
Feldman: This is awesome. What do young people do with phones? It's just amazing. I was thinking you'd have to glue some sensors onto people's ankles, knees, and hips, and then really capture some minute data, but you captured so much just with the phone itself.
Ogdie: Phones now have sensitive goniometers, so you can get motion and accelerometers, and so all those pieces together can capture really large amounts of data, not to mention the fantastic cameras. And then if you combine that with an Apple watch, let's say, you can also get reasonable data on sleep, heart rate, et cetera.
You're getting so much information that we can potentially use to understand health outcomes, whether that be looking at cardiovascular risk in a psoriasis patient or looking at psoriatic arthritis risk vs how it's happening or just general health.
Feldman: Yeah, I've wondered if that Apple Watch could tell me who's scratching and when they're scratching as a best continuous measure of skin disease severity for something like psoriasis or eczema.
Ogdie: Exactly. I think Lourdes Perez-Chada, who's one of the authors on this paper — I think she was the first author, maybe, I can't remember for sure — she's testing some of those things with different digital devices as well and how it interrupts sleep.
Feldman: This paper validated this tool. So when I see patients with psoriasis. I think I get most of the information I need for treatment purposes by asking them, how are you doing? You know, you're doing great. Yeah, I'm doing great. Well, man, I'm not going to change anything. Oh, I'm not doing so well; then I'm going to change the therapy.
That's about it. Do I need a validated tool in my practice? Or do I need a validated tool for research?
Odgie: Initially this would be for research and for patient interest. I'm very interested in all my digital biomarkers. I have my Oura ring and my Apple Watch and so on, so I like to watch my own data and that's part of what we anticipate that people using this will like to watch their own movement, for example.
In clinical practice, there's this problem called the 5,000 hours problem or this concept that we're only seeing short segments of time in the patient's life. And so we don't see all the other things that are happening. In another study that we're doing, we're asking patients at intervals questions about their psoriasis, and then if they're flaring, and 50% of the time, people say they're flaring. So there may be some variability in how people are doing. So this is meant to kind of understand better what the variability is in how people are doing. For example, also, in psoriatic arthritis in particular, I think also in psoriasis, the faster someone responds, the better they feel, and sometimes will take trade-offs for taking a longer time to get to the same level of disease activity.
One of the things that we want to test with this app is whether we can see the trajectories over time and how people respond to therapy and get some insights about what may be better for certain people, or how certain therapies may work differently for certain people.
Feldman: The new measures are very interesting. What measures have we usually used to measure disease? My guess is you rheumatologists spend more time with the patient: spend more time recording useless, validated, quantitative measures than I do.
Ogdie: We do record a lot of things. We have a paper, I think we had an abstract recently, or we'll have an actual and paper that demonstrates that, you know, as specialists in psoriatic arthritis, we are better at capturing lots of different kinds of data than a general rheumatologist seeing all different kinds of things.
But then the question is actually, does it matter or not? That's a separate question for a separate day. But we do capture all kinds of things in our specialty clinic, including tendered, swollen joint counts, and these number of enthesis that are tender, swollen — really, dactylitis. We capture psoriasis body surface area and the Physician Global Assessment erythema scale. And then we also capture nail disease. So we're writing down all of these things, and they do make a difference in terms of thinking about what the patient has active right now and how we want to select a therapy.
I think one of the other good things about capturing data objectively and repeatedly capturing it objectively is that it keeps us on track toward a goal and makes sure that we know whether things are going up or down for the patient. For example, when a patient comes in, you ask them how they're doing, and they say I'm doing quite well. You know, I feel really good today. But you see that the score for the RAPID3 (Routine Assessment of Patient Index Data 3) index, which we use in our clinic, has gone up, pretty substantially from the last time they're in, you know, that's now a new reference point. So they're feeling pretty good today compared to how they felt 2 weeks ago, but they're still not doing very well.
And so, having that objective data point helps you make a comprehensive treatment plan and then put everything into perspective. So I think in some ways it gives it a full perspective.
Feldman: I think the way people see things is colored by the context in which they see it, and you're dealing with a disease that can cause scarring and permanent deformity and disability, and I'm dealing with a disease with skin involvement, which is where I say 'Hey, you clear up the skin disease, the skin's back to normal.'
Because the skin is like one of the best organs in your body, maybe the best, well, most important organ in your body.
Ogdie: Well, I mean we can contend whether that's true or not, but it is the largest, I suppose. And also it is, in my opinion, one of the easiest to assess, because you can see it. So we're often assessing a lot of different things you can't see, which is challenging.
Feldman: After reading your article, I thought about if you've just had a little stand for the phone on a rod that goes around you and goes up as it goes around and it would give you a helical scan of the body and not only could you tell if there were any moles that had changed, you'd really fully capture their psoriasis disease severity.
Do you just have people take a few pictures?
Ogdie: Yeah, we just have them take a picture of a representative plaque, and that kind of follows what's done in clinical trials in general. There's a target lesion often, so essentially we're having them pick a target lesion that we can kind of continue to follow over time as they're on therapy.
Feldman: Does the phone in your system capture the full body surface area?
Ogdie: The patient kind of does a little drawing. And there's a little video you can make. So over time, they will continue to add different days when they're measuring their body surface area, and then it will do a little video. It can spin the body, and it can also show over time how that's changing.
It's kind of nice, I think. Patients like the idea that they can see their body surface area psoriasis going down, and every time point when you're putting something and you know only what is today and there's good data that says that maybe your 7-day recall is pretty good, your 4-week a recall is okay, but beyond that, it's not so good.
Like almost logged recall, so you don't have to remember back then. And that shows you where you've been and where you're going.
Feldman: Yeah. Do you ask them about adherence to their treatments?
Ogdie: Not in this one, in a different study. I think there is a single question about that. If I had not had to go back and remember, it's not something we put in the paper, but, there may be a question about treatment and treatment satisfaction. In other studies, we do do that. We ask about treatment satisfaction, treatment adherence. We actually just finished a dietary trial where we were asking about adherence daily so that they would have, it's almost like a feedback mechanism.
Feldman: Yeah. Excellent. Well, they're probably lying.
Ogdie: Well, it'll be interesting to see — we're analyzing that data now because we have actual urine biomarkers, for example, and other things.
Feldman: Oh, nice. So you're measuring adherence by your environment now. That's cool.
Ogdie: That'll be fun to analyze.
Feldman: Yeah, and lying may not be the right word, because like you say, they don't remember — recall is a factor, and recalling adherence to pills or whatever treatments can be tough.
Ogdie: It is. Yeah. In one of our other trials, we have a pragmatic trial where we're randomizing patients to one of three treatments, and they have to log when they receive it. And then they log again when they take it, because otherwise, exactly: No one remembers what day they took their last biologic dose, or many patients don't remember that.
Feldman: Are you using digital tools or AI in other ways?
Ogdie: One of our studies is examining predictors of developing axial spondyloarthritis (AxSpa) in the [electronic health records]. So there we're using text-based tools. So, picking up people talking about Achilles tendon pain, for example, in a primary care appointment, and trying to figure out how long did that happen before they developed their AxSpa?
And if there was a series of features that were being mentioned, should someone have thought about that diagnosis at a certain time? And is there a way to identify or flag those patients?
Other ways we're using AI or digital tools or in large databases, again using coded data. So ICD codes, for example, or prescriptions to say, could we predict when someone's developing psoriatic arthritis or AxSpa? One of my colleagues is doing vasculitis like that. All of these things though, require us to utilize existing data. And one of the main problems with existing data is that there's a lot missing from existing data.
So you only get the snapshot that you're looking at. It's like picking what you want to stare at, and then you just stare at that. So we're missing so many other things like the fact that someone's got wrist pain and so they're on the web looking for wrist pain. What is my wrist pain? Or they're ordering something from Amazon to go over their knee because they're having knee pain.
And all of those things could be so helpful in understanding whether someone's developing psoriatic arthritis and it's not going be in a code anywhere. So AI is only as helpful as the data you plug into it, and that's one of the challenges today.
Feldman: Oh my, that reminds me of the story that I think Target sent some family.
Ogdie: Yes. Pregnancy materials and that. Yeah, exactly.
Feldman: Before the family knew they were pregnant. Yeah, so I wonder if Amazon has the data now to tell us who's developing psoriatic arthritis before we or the patient even knows.
Ogdie: Well, they probably do, they just don't have the outcome. So in this case, they have maybe all the exposures we might want among Prime members, for example. If you're buying things from all different places, it wouldn't be as helpful. But you know, they don't know the patient has disease X, Y, or Z.
So, if we could merge those two things, maybe we could come up with lots of interesting insights. But, you know, then we have to talk about the balance between privacy and good prediction models.
Feldman: I mean, if I were an insurer, I'd want to work with Google and find out who had joint pain ahead of time, so I would know not to sell them a policy that covered biologics.
Ogdie: Yeah. There's some definite dangers in the use of these tools as well. Not in the way we've designed our current tools, so to be clear, but in tools that could involve more, other extraneous information. We can learn a lot. We can probably do a lot of good at the same time.
Just like all the conversations around AI in, in the world. There's the dark side of that, which is insurance companies or someone blocking them from getting assistance for something.
Feldman: This iPhone I have is not the newest model, but the camera on it probably sees better than I do. Its memory is certainly better than mine at my age, and it can be trained on more skin diseases and things than I can be trained on. So I'm wondering, you know, I've been thinking about taking my wife on a long cruise to Antarctica, which would mean being out of the clinic for a prolonged period, but I'm thinking pretty soon it won't matter because they're not going to need me.
Ogdie: Well, I still believe that we are always going to have a role and an important role. So it's just like the, you know, first of all, for rheumatology, we need all the help we can get. Because we just don't have enough visits for everybody. I think that and we have a major workforce issue. And that workforce issue is going to continue to be a problem even more so in the next 10 years.
And I know dermatology has not enough visits for everyone too. So if we could reduce some of the burden or a allow patients to have more self-efficacy and kind of control parts of their own disease and help them with self-management, I think we'd be doing a service to all of, us because then you wouldn't get as many phone calls when you're in Antarctica that you have to answer.
I think that's where we want to take some of these tools. How can you help the patient take care of themselves? And then the doctor still has a role and they have to help interpret in light of the information, in light of who the patient is. But maybe we can take some of that off the doctor, so we can use the doctors for the more complex pieces.
Feldman: Yeah, I don't know. I've been using Claude to help me write thank-you notes and things, and Claude knows everything.
Ogdie: It knows a lot. It's interesting though, if you ask some of these things about psoriatic arthritis, it takes what's on the web and a lot of what's on the web is patient blogs. For example, patient blogs of psoriatic arthritis and AxSpa are dominated by like fibromyalgia type discussions, and so what they know of our diseases is more about pain, and so it's not as much about some of the other pieces. So there's a biased kind of idea of what is in the general lexicon.
Feldman: Yeah, It will be interesting to see what would happen if you asked some patients' questions, typical questions. I'm in pain, I'm in these areas to the doctor versus to the AI, and see who gives more empathetic, caring, accurate responses.
Ogdie: No, that's a separate thing. That is very true. You can definitely train it to give a certain type of voice.
Feldman: Yeah, well today we've had Dr Alexis Ogdie discussing digital assessment tools. We've learned that you can remote monitor with just your phone in many, many ways. These digital tools may give patients another quantum leap forward in terms of the questions that they can answer for themselves.
I hope you all will come back for episode six, where we'll conclude this season with Dr April Armstrong discussing new and novel topical therapies for psoriasis. Thank you so much for joining us. This is Dr Steve Feldman for Medscape's Psoriasis InDiscussion .
Listen to additional seasons of this podcast.
Clinical Validation of Digitally Acquired Clinical Data and Machine Learning Models for Remote Measurement of Psoriasis and Psoriatic Arthritis: A Proof-of-Concept Study
Psorcast
Psoriasis and Psoriatic Arthritis Clinics Multicenter Advancement Network Consortium (PPACMAN) 2021 Annual Meeting Proceedings
Sleep Problems in Patients With Psoriatic Arthritis: A Systematic Literature Review and Metaanalysis
The Impact of Dietary Interventions in Psoriatic Arthritis
