touchRESPIRATORY coverage of AAAAI 2025:
Artificial intelligence is no longer a futuristic concept in respiratory care — it’s already here, reshaping how we diagnose and manage asthma. In this insightful interview, Dr Alan Kaplan shares how AI is helping bridge long-standing gaps in asthma diagnosis, adherence, and disease control.
From apps that can tell the difference between asthma and COPD, to smart inhalers that track medication use and predict exacerbations, Dr Kaplan explores the real-world potential of AI — and the challenges that come with it. He also weighs in on whether AI scribe software can enhance the patient–physician relationship, and what excites him most about the road ahead.
Questions:
1. How is AI changing asthma diagnosis and management?
2. Does AI scribe software have the opportunity to enhance physician-patient interactions?
3. What are the main challenges in using AI tools such as smart inhalers?
4. Can AI-based asthma management tools help bridge the gaps in adherence and disease control?
5. Looking to the future, what excites you most about AI-driven advancements?
Associated Abstract: Use of AI in Asthma Management and Treatment Decisions. Presented at 2025 AAAAI/WAO Joint Congress, San Diego, CA, USA. February 28-March 3rd 2025.
Disclosures: Dr Kaplan has nothing to disclose in relation to this video interview.
This content has been developed independently by Touch Medical Media for touchRESPIRATORY. It is not affiliated with AAAAI. Unapproved products or unapproved uses of approved products may be discussed by the faculty; these situations may reflect the approval status in one or more jurisdictions. No endorsement of unapproved products or unapproved uses is either made or implied by mention of these products or uses by Touch Medical Media or any sponsor. Views expressed are the speaker’s own and do not necessarily reflect the views of Touch Medical Media.
Transcript:
1. How is AI changing asthma diagnosis and management?
Well, the big picture here is that artificial intelligence will hopefully assist clinicians, patients, and their families.
We have still a huge issue with asthma, in the United States where I presented this talk, and there’s over thirty million pea, people with asthma, one in eleven children. They’re spending eighty billion dollars plus on asthma, and asthma control is still an issue. There’s still three thousand five hundred deaths annually in the United States.They have tons of missed school, missed work, missed life opportunities, decreased physical activity, overuse of short acting beta agonists. So with all that, we have we have some real big issues to deal with. In terms of what we can do about it, let’s break it down into a few things. First of all, diagnosis.
So in terms of diagnosis, what can AI help?
Well, we created an app called ACDC or asthma COPD differentiation, app. And with this app created with, machine learning, we can only put twelve questions with a microsporometry to actually inform clinicians about the difference in asthma COPD, which is a common clinical question.
And it was found to be efficient and actually more efficient than clinicians.
And because computer learning does have the opportunity to bring in so much data and make decisions with things that we might find extraneous.The downside to that, of course, is anything you do, you train with the system. And if the system isn’t given enough information to train properly, you can get errors, and I’ll come back to that.
Another there’s another piece of work done. We showed some European pulmonologists, and they actually compared their ability to read spirometry to an artificial intelligence program. And once again, the AI program was was a bit better. So it doesn’t mean that AI outperforms clinicians, but it may out of outperform groups of clinicians in terms of large amounts who have different interests.
Moving into management. So overall, I talked about the lack of control and some of the, the, issues in that United States. Of course, the same around the world.
And I think we can look at some of the causes of lack and control that AI can help. These include things like adherence, inhaler technique, as there’s usually the big issues as well as diagnosis. So diagnosis would be the first thing and then, and then adherence and inhaler technique.
The other issues would be things like comorbidities and triggers, and some people will just have more severe asthma.
So what can we do with AI? Well, smart inhalers are are these things that add on to inhalers or part of inhalers that include things also like apps. And these smart inhalers have the ability to measure adherence count count doses, but also include things like reminders. And with the apps, it can include questionnaires.
And all kinds of other data can be added on to these apps overall to help things like, the air quality in the area and things like that. And with all of that, it’s been reviewed to see if it makes a difference. And Cochrane reviewed this a few years ago, and it showed that these these smart inhalers do improve adherence control and can decrease exacerbations.
So there’s a lot of potential benefit for that.
So we we have other apps that can even include things like phenol or fractionated cell nitric oxide or CRP, which is a measurement of inflammation. And the more information you give the app, the better it’s gonna be able to predict. And some of these apps can predict exacerbations by measuring increased Saba use and and and decreasing inspiratory flow, which is predicting exacerbation.
There’s another piece of electronic equipment called electronic stethoscope that’s given to parents of children who just had exacerbation.
And by listening to by listening with this electronic stethoscope, they can actually hear early wheezes and, again, predicting exacerbations.
Another one is on a on an iPhone where you actually it counts cough. And by counting cough, it can again predict exacerbations developing as they occur.
We’ve got other things as well to help with technology with digital spacers. So spacers are holding chambers. You plug motemeter dose inhalers in, and it allows better better, inhalation properties of the medication into the patient. Now we have digital spacers that can actually improve technique of the spacers as well.
Other things that are done maybe more more about the patient physician interaction, things like scribing. So what happens there is in the background, a computer is listening to the patient physician interaction and removing extraneous data and pulling out based on the algorithms created with the scribe, the medically necessary data to make that to happen.
That can actually allow, and then the scribe writes it up. So that actually can be quite efficient. It can be very helpful. Also, what it does is it allows instead of me sitting on the computer, typing over here while we’re you’re talking to me, I can look right at you and listen, recognizing the data is being captured by the scribe. So these are the kind of things that are happening out there. Physicians are already starting to use this technology, and it’s gonna improve the physician patient relationship because we can actually have that more direct one on one.
Those scribe notes can then be used can then be used as even as even handouts for the patient afterwards.
So with all this kind of stuff, including things like software that can monitor our inbox as patients send in questions and can write initial answer, we can’t just trust it as an answer. We still have to read it and check it, but it’s gonna save some time because it’s certainly faster to edit an answer than to create it from fresh from scratch.
2. Does AI scribe software have the opportunity to enhance physician-patient interactions?
Well, I think that’s the opportunity. A few things. It certainly has the opportunity to potentially save a little bit of time. You can be a little more efficient in your appointment by recognizing you don’t have to be typing everything down or writing everything down.
The scribe is gonna do that for you. That being said, scribes have been known to make mistakes. They are trained on data that they’re trained on. A lot of that training has been done on things like social media, and it’s only more recently that some of the big companies like ChatGPT are using more medical textbooks and medical issues so they can actually get the right information.
There has been, some work actually where where I saw some work from a study from where, the clinician said the problem’s got to do with your hand, foot, and your mouth, and the scribe read hand, foot, and mouth disease. Or the clinician said, you need you need to get your prostate exam done, and the scribe said prostate exam done.
So the message here is that it’s really important as a clinician that you need to read those notes and ensure they’re accurate. You need to read those at inbox responses I mentioned and make sure it’s accurate because otherwise, you may have incorrect notes and incorrect information going out to to patients. And that, of course, has medical legal issues as well as quality of care issues.
So it has opportunities to benefit. They did a a measurement in Ontario about this to see whether it was actually gonna save time, And it actually did save some time and save some fatigue in the in the physician in terms of typing and so on at the same time. But, again, it’s not gonna save as much as you think if you don’t read it yourself. And if you do if you don’t read yourself, you’re taking a chance of missing something important.
3. What are the main challenges in using AI tools such as smart inhalers?
Well, like everything, it’s gonna time is gonna be the big one. You have to have time to learn how to use it, time to monitor it, time to implement it, time to train for it. I’ve already talked about things like accuracy of scribe notes and things like that, how we can get into trouble. Privacy.
Who’s listening? So if that AI that I mentioned that’s counting the costs, excuse me, like that, what else is it listening to? Is there any other data that’s gonna be picking up? And so who’s gonna have that data? Who’s gonna be monitoring that data?
And overall, technology can be quite overwhelming to some people. We did a program called magnify, which is a COPD app, for COPD app. And the question was, is patients with COPD who are a bit older, are they gonna be overwhelmed by the technology? It was interesting that, actually, it it wasn’t overwhelming to them. They were able to manage, but some of them just still just couldn’t be bothered doing it because they didn’t wanna bother about the entire process because it it does require some work by the patient.
I’ve heard of some fear of being watched. You know, it’s, you know, right right now, there’s people out there that believe that, there’s nanotechnology and flu shots monitoring everything you do, so they don’t wanna get a flu shot. Hate to tell you, but if you got an iPhone, you’re being monitored all the time. You know, when you when’s the last time you picked up and opened up Instagram and you got an ad for something you were talking about two days earlier? So these are the things that are happening.
Who’s over where where’s the oversight in all this? This? And, if something does happen incorrectly based on, the information you get, who’s responsible for it? Who’s gonna pay for it? Who’s gonna pay for these things like smart smart inhalers?
One of the concerns is it’s gonna be a push or a pull technology. What I mean by that is if you’re using the smart inhaler and there’s reams and reams of data coming in, how many doses they took, how the peak inspiratory flow was, they’re also reading in their control questionnaires.
And that’s coming in. Am I supposed to be monitoring this on every single one of my patients every single day?
That’s that’s an it’s an untenable situation. That has to be monitored by a computer program with set algorithms to set off alarms that we can then act upon.
Or it’s just a pull technology. So when the patient comes in, we can then review a summary of that issue and and and review it as part of the control parameters. Because when we’re with the patient, we’re with them for fifteen minutes every three months or so. You Now that’s a very tiny amount of the total amount of time that we’re that the patient is living. So lots of ways to do this, but, there’s still lots of concerns about it.
4. Can AI-based asthma management tools help bridge the gaps in adherence and disease control?
Today, I can help bridge some of the gaps because time is time and communication are some of the biggest issues we have in medicine. I think, as I said before, it’s gonna allow it more direct patient physician interaction, and it can monitor and it can predict.
So it’s really gonna help us analyze analyze what’s happening to the patient on a day to day basis and and and prevent them from getting into trouble by assisting them by early intervention, early warning.
5. Looking to the future, what excites you most about AI-driven advancements?
In the future, well, I think overall that we want these the AI to be a tool to help us manage the patient, not in itself be the endpoint.
It’s not to decide diagnosis and management itself, but it’s to assist us in that. The concept here is that we want AI to be able to assist us in gathering data, but it still requires the human judgment to use that data. And we can only gather so much data at one time, but a computer with its with its computing technology, it can it can process so much more data than a human can. And therefore, it can give us predictive models that then we can act upon.
So that’s where the that’s where it’s gonna be. Human judgment still has to be in there. And just as a note, you know, I’ve asked something like chat GBT, you know, can can chat GBT make clinical decisions? And it will say no, Chat GBT should not make clinical decisions.
Should Chat GBT replace physician judgment? If you ask Chat GBT, it’ll say no. It should not replace physician judgment.
So we have to work with it. So the concept I think here is AI will not replace physicians, but physicians who use AI might well end up replacing physicians who don’t use it. So overall, know its limitations, and please double check its outputs because I think otherwise a lot of harm could happen.
Interviewer/Editor: Nicky Cartridge
Cite: Kaplan A. Is AI the future of decision-making in asthma diagnosis and treatment? touchRESPIRATORY. March 26, 2025.
